JP2001053747A - Shaping method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To uniformly arrange whole groups in terms of structure by collecting the groups whose bands are the same, arranging the groups from the group of high priority and the group of the large band, calculating ideal arrangement slot positions for respective groups, arranging group elements to the positions and retrieving an alternate slot position at random. SOLUTION: In GID arrangement where a group identifier (GID) is written into a GID arrangement table, the groups of the same bands are collected and the groups are arranged from the group of the large band with base offsets selected at random from whole slots as a reference. A start offset in the group and GID in one group are uniformly dispersed. When intra-group GID cannot be arranged in a permitted CDV value (deviation from ideal position), the start offset is changed and it is reset. When it cannot be reset even if the permitted number of times is changed, an integerizing method and the permitted CDV value at the time of calculating a GID interval are changed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a shaping method implemented in a switching device of an ATM network node.

[0002]

2. Description of the Related Art FIG. 77 is a diagram showing a conventional example shown in, for example, IEICE Technical Report SSE98-193 (January 1999) Iwamura, Kobayashi, Horie, Fujita, Ashi "A Study on Bandwidth Assignment Method of GFR Processing Circuit". It is a figure which shows the shaping method of. In the figure, GIDTBL is a group identifier arrangement table, ADR is a GID address, and CALP
Is a calculated ideal arrangement position of GID, CALD is an ideal arrangement interval of GID, and RD is a correlation calculation range.

[0005] In FIG. 77, an extract and enlargement of seven slots before and eight slots after the ideal arrangement position CALP are shown at the bottom of the figure, and numbers 1 to 16 are assigned to each slot. Slot 5 to slot 12 are acceptable CDs
This is an allowable CDV range in the case of a V value of 7, and for each slot, a correlation ideal range RD5 to RD12 is defined for every nine slots.

FIG. 78 is a diagram showing a procedure for arranging group elements in one group in the conventional shaping method. In the process trunc_s3_past_s3 shown in the figure, the ideal arrangement interval CALD is calculated by the total number of slots / the number of group elements GPCR, and based on this, the ideal arrangement position C
ALP is required. Assuming that the ideal CDV value is 7 within the allowable CDV range for each ideal arrangement position CALP,
When the front 3 slots to the rear 4 slots, that is, when the ideal position is set to the slot 8, the weighted correlation of the correlation calculation range RD5 to RD12 is calculated for each slot from the slot 5 to the slot 12, so that the arrangement density is as uniform as possible. Select the placement position.

Further, the GID table GID
With respect to the arrangement position of the ideal position CALP in the TBL, 10 positions are selected and tried within the range of the ideal arrangement interval CALD, and the positions where the overall arrangement density becomes as uniform as possible,
That is, a position where the arranged slots in the GID table GIDTBL are made as equal as possible is selected, and the GID is arranged.

In the above-described conventional shaping method, various kinds of complicated virtual arrangement processing are tried for the GID arrangement, and the results are compared to select an optimal arrangement position. As shown in the procedure of FIG. 78, this method imposes virtual arrangement processing including multiple loop processing.

[0007]

In the method shown in FIGS. 77 and 78, the allowable CDV value is 7 or less, but the allowable CDV value is not more than 7.
As the value increases, trunc_s3_past_s4
The number of loops from to trunc_s3_past_s7 increases in proportion to the permissible CDV value, and it is necessary to go through virtual placement processing by multiple loop processing even when the ratio of placement required group elements to all slots in the GID table is small. There were challenges.

[0008] In this method, the CDV at the time of arrangement is
Since the structure is such that the maximum value is set so as not to exceed the allowable CDV value, there is a problem that even if the arrangement can be made with a small number of CDVs, the arrangement is made at a position with a large CDV value.

[0009] This method is also suitable for the ideal arrangement position CALP.
Is calculated by dividing the quotient of an indivisible integer division by a simple integer (ideal interval CALD = total number of integer slots /
There is a problem that the calculation is performed at a constant value using an integer number of group elements (GPCR), and there is no consideration for the deviation from the ideal position based on the real number calculation and the conversion to an integer.

The present invention has been made to solve the above-mentioned problems, and a first object is to arrange all groups structurally. The second purpose is
The ideal arrangement positions are distributed as evenly as possible among all the slots. A third object is to select an appropriate alternative position when it cannot be arranged at the ideal arrangement position. A fourth object is to enable an accurate CDV value to be calculated. A fifth object is to obtain a random number sequence having a limited period and range.

[0011]

According to the present invention, there is provided a shaping method, comprising the steps of: defining a group of one or a plurality of groups; In a distributed arrangement, a grouping process for arranging the groups having the same band at once, a priority arrangement process for arranging the groups having the higher priority, a descending order process for arranging the groups having the larger band, An ideal position calculation process for calculating an ideal arrangement slot position (hereinafter, a slot position is referred to as a position), a group element arrangement process for arranging a group element at the calculated ideal arrangement position, And a substitute position search process for searching for a substitute slot in the case where the ideal position calculation process is performed.
When uniformly arranging a predetermined number of the groups in the matrix table, an integer interval calculation process for calculating two types of integer arrangement intervals, and the ideal when uniformly arranging a predetermined number of groups in the matrix table In the real number interval calculation process of calculating the arrangement position by real number calculation and converting it to an integer, in the real number interval calculation process, when converting the value obtained by the real number calculation to an integer, add a bias and then perform an integer conversion operation process A bias addition process, an interval mixing method changing process for changing the selection order of the two types of integer arrangement intervals, and a group in which the ideal arrangement position of each group in the same band group is evenly distributed within the integer arrangement interval. Ideal position equal distribution processing, and within-group start position equal distribution processing for equally distributing the arrangement start positions of the same band group for each group in the matrix table. An ideal arrangement position calculated by the real number interval calculation process and the intra-group ideal position uniform distribution process; and a matrix coordinate storage process of storing the distribution start position calculated by the intra-group start position uniform distribution process as matrix coordinates; The arrangement start positions of the different groups are randomly distributed in the matrix table, and the start position random distribution processing and the equal distribution of group elements in the same band group are performed using the integer interval calculation processing and the real number interval calculation processing. And an adjacent arrangement prohibition process for prohibiting the arrangement of elements of the same group in adjacent slots. The alternative position search process observes the slot arrangement situation around the ideal arrangement position. A pattern matching process for obtaining an optimal arrangement position by comparing with an optimal arrangement position pattern stored in advance, An arrangement ratio dependent pattern matching process for changing the optimal arrangement position pattern depending on the ratio of the arranged slots in the body table; a neighborhood random search process for randomly searching for a nearby slot at the ideal arrangement position; Random search processing for randomly searching the matrix start position in the matrix table, and randomly searching the matrix table for an arrangement start position reference point of the same band group in the sky of a predetermined continuous slot in the matrix table. And a continuous random search process.

According to the shaping method of the present invention, in the bias adding process in the ideal position calculating process, the decimal bias adding process in which the added bias value is equal to or more than zero and less than 1 and the interval mixing method changing process in the ideal position calculating process are performed. The bias value is changed by the decimal bias adding process and the group element is reset, and the order of the bias value to be changed in the bias changing process is set to a value of zero or more and less than one. It is provided with a sequence of integer values and a bias value change pattern storing process for storing the total number of types of bias values as integer values by multiplying the bias values by the total number of types of bias values.

According to the shaping method of the present invention, in the real number interval calculation process in the ideal position calculation process, 32
A first process of dividing a parameter by a divisor using a fixed-point variable having an integer of bits or less to obtain a first quotient and a first remainder, and shifting the first remainder to the left by a predetermined number of bits A second process of multiplying (2 by a predetermined number of bits) and a third process of dividing the left-shifted first remainder by the divisor to obtain a second quotient and a second remainder. Multiplying the second quotient by a first integer equal to or greater than 1 and equal to or less than the divisor to obtain a first variable; and multiplying the second remainder by the first integer; A fifth process of dividing by the divisor to obtain a second variable;
And the first surplus answer is right-shifted by the predetermined number of bits (divided by 2 a predetermined number of times). , Second
A seventh process for obtaining a surplus answer of the above, an eighth process for multiplying the first quotient by the first integer to obtain a quotient answer, the second surplus answer and the quotient answer A ninth process for obtaining an answer, and a loop process for changing the first integer in a range from 1 or more to the divisor or less and repeating the fourth to ninth processes. It is provided with.

[0014] The shaping method according to the present invention includes a start position random distribution process in an ideal position calculation process, a neighborhood random search process in an alternative position search process,
In the whole random search process and the whole continuous random search process, a random number period represented by a power of 2 is used as an argument, a different variable is provided for each period, and a random number initialization process for initializing a variable for each period is performed. A process of updating each variable with a linear function, a random number generation period control process of randomly generating each integer between zero and a maximum destination once or twice when a random number is generated for one period, and It is provided with a cycle selection process for selecting a minimum power-of-two cycle that does not exceed, and a random-number function calling process for calling a random-number function of the selected power-of-two cycle.

In the shaping method according to the present invention, in a pattern matching process in an alternative position search process, an optimum layout position pattern creation process performed in advance, an optimum layout position pattern storage process performed in advance,
An optimal arrangement position pattern reference process to be performed at the time of pattern matching is provided. When the optimal arrangement position pattern creation process includes empty positions suitable for the same degree, all of the arrangement candidates are arranged. When there is a vacant position where continuation does not occur, a process of setting the vacant position as a placement candidate;
When there is no empty position where the arranged continuations do not occur, a process of setting the position having the smallest number of arranged continuations as an arrangement candidate, wherein the pattern stored in the optimal arrangement position pattern storage process is an ideal arrangement The arrangement situation of the slots around the position is represented by a continuous sequence of 1s and 0s, where the arranged slots are 1 and the empty slots are 0, and this is regarded as a binary sequence and converted to a decimal number, and an optimal arrangement position. The number of arrangement candidates in the pattern and the arrangement candidate positions in the optimal arrangement position pattern are represented by relative arrangement positions with respect to the ideal arrangement position, and the arrangement candidate positions are listed. A process of investigating the arrangement status of peripheral slots and creating an arrangement status pattern;
A process of referring to the pattern stored in the optimal arrangement position pattern storage process, and a process of selecting one of the arrangement candidate positions by using a random number when the number of arrangement candidates stored in the optimal arrangement position pattern storage process is plural. It is provided with.

[0016] In the shaping method according to the present invention, in the pattern matching process and the near random search process in the alternative position search process, the alternative position actually located when the ideal position cannot be located is shifted to the right side ( The borrowing, which is a variable representing the difference in the number of slots (the difference between the address of the alternative position and the address of the ideal position) when the slot is arranged at the larger slot address, is added to the calculated ideal location position address, and the alternative position address is subtracted. The index calculation processing for obtaining the index by performing the calculation, and when the index is positive or zero, the difference in the number of slots when the alternative position is arranged on the left side (smaller slot address) in the matrix table from the ideal position ( The index indicating the narrowing indicating the difference between the address of the alternative position and the address of the ideal position), and zeroing the new borrowing, In the case where is negative, the narrowing is set to zero, the new borrowing is performed as an absolute value of the index, and the borrowing and updating process is performed, and the new borrowing is added to the integrated value of the previous borrowing. It is a thing.

[0017] A shaping method according to the present invention is a method for adjusting a transmission interval of fixed-length cells transmitted based on an output band determined for each connection of one or a plurality of connections. A process of rearranging groups in descending order of bandwidth values required for a connection group (hereinafter collectively referred to as a group) configured as a set of connections, and a fixed length for a group to be processed in bandwidth value order A process of calculating an ideal transmission interval at the time of transmitting a cell, and, on a matrix table which is composed of a plurality of slots each having an address and defines an output order of fixed-length cells, an arrangement start position of a processing target group is determined. A process of sequentially determining an ideal arrangement position for all the arrangement positions including,
A process of determining whether a group to be processed can be arranged at the ideal arrangement position, and a process of arranging a group to be processed at the ideal arrangement position if the arrangement is possible,
When the arrangement is impossible, when the total band value already arranged is smaller than the judgment value, the processing target is preferentially given to the odd addresses (or even addresses) within the allowable CDV value centered on the ideal arrangement position. And arranging a group, and arranging the processing target group at an arbitrary address within a range of the allowable CDV value when the allocated band total value is larger than the determination value.

A shaping method according to the present invention is a method for adjusting a transmission interval of fixed-length cells transmitted based on an output band defined for each connection of one or a plurality of connections. A process of rearranging groups in descending order of bandwidth values required for a connection group (hereinafter collectively referred to as a group) configured as a set of connections, and a fixed length for a group to be processed in bandwidth value order A process of calculating an ideal transmission interval at the time of transmitting a cell, and, on an output table which is composed of a plurality of slots each having an address and defines the output order of fixed-length cells, sets an arrangement start position for a processing target group. A process of sequentially determining an ideal arrangement position for all the arrangement positions including,
If the total value of the allocated bandwidth is smaller than the judgment value,
If the group to be processed is preferentially arranged at odd addresses (or even addresses) within the allowable CDV value centered on the ideal arrangement position, and if the total allocated band value is larger than the determination value, the ideal Allowable CD centering on the placement position
And arranging a group to be processed at an arbitrary address within the range of the V value.

In the shaping method according to the present invention, the judgment value to be compared with the total band value is given as a half of the total output band of the matrix table.

In the shaping method according to the present invention, when the group to be processed is preferentially arranged at odd addresses (or even addresses) within the allowable CDV value, the remainder obtained by dividing by a predetermined number becomes a specific numerical value. A group to be processed is arranged at a different address.

[0021]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described below. Embodiment 1 FIG. First, an example of group identifier arrangement using a shaping method according to an embodiment of the present invention will be described. FIG. 1 shows an example of a group identifier arrangement table. In the figure, GID is a group identifier, S
LOT is a slot which is a memory space for containing a GID,
ADR is the address of the slot SLOT, GIDTBL is the GID address 11980 of ADR from 0 to 11979
It is a group identifier (GID) arrangement table composed of a number of slots SLOT.

The band-like GID arrangement table (G
The ID table band) is logically connected at both ends, and is equivalent to the annular GID arrangement table (GID table ring) shown in FIG. FIG. 3 summarizes the terms shown in FIGS. 1 and 2.

Next, prior to a specific description of the present embodiment, the background and the positioning of the need to perform GID arrangement using the GID arrangement table in this embodiment will be described.

With the rapid development of the Internet and the like, ATM (Asynchronous Transf)
er Mode (asynchronous transfer mode), there is a demand for a method of realizing inexpensive data communication. Among them, ATM service class GFR (Guara)
attention is being paid to AT
M Forum and ITU-T (International Telecommunication Union) are standardizing on ATMForum Technology
al Committee Traffic Mana
Geometry Working Group: “Tra
fic Management Working G
group Living List (ATM Foru
m / LTD-TM-01.04 to 01.0
6), “Jullyto Dec. 1997, and IT
U.T. Recommendation I. 371 LivingList, Gen
eva, Jun. Documents such as 1998 have been published.

In GFR, the minimum bandwidth of each connection (mCR; minimum Cell R)
ate), and if there is room, the maximum speed (P
A surplus bandwidth exceeding mCR is allocated to each connection within a range not exceeding CR (Peak Cell Rate). When allocating a bandwidth to each connection, a multiplexed CDV (Cell Delay Variation) is used.
n; cell delay variation), fairness, and the like need to be considered, and GFR bandwidth allocation is an important technical issue.

FIG. 4 is a conceptual diagram of band allocation in GFR. In the figure, HYWAY is a highway of a communication channel,
CONNCTN indicates an ATM connection. Here, a highway HYWAY of 1.2 Gb / s (precisely 1.198 Gb / s) accommodates a maximum of 2048 connections CONNCTN. Connection CONNC
When accommodating the TN in the highway HYWAY, a surplus bandwidth is shared between a plurality of connections. The bandwidth sharing unit BAN performs the bandwidth allocation for sharing the bandwidth between the connections.
DM. At this time, bandwidth sharing is not performed collectively for all connections, and connections are limited to a maximum of 512 group GRs.
Divided into OUPs and performed within each group. That is, G
FR bandwidth allocation is performed in two stages: a connection CONNECTN unit and a group GROUP unit.

In the band sharing unit BANDM, as a method of allocating the band within the group GROUP to the connection CONNCTN, a WRR (Weighted Round) is used.
Robin; round robin with load) method and common F
An IFO buffer scheme is under study.

The WRR method is described in, for example, M. Kateven
is, S. Sidiropoulos, C.I. Courc
obettis, "Weighted Round-R
obin cell multiplexing in
a general-purpose ATM sw
hitch chip, "Journal on Sel
cted Areas in Comm. , Vol.
9, no. 8, pp. 1265-1279, Oct. 1
Connection CONNECT in the manner described in 991
A buffer is provided for each N, and cells of each connection are read out sequentially with a load.

The load for each connection is calculated by the identifier (CID: Connection Identifier) of each connection.
Identifier) is held in the connection identifier arrangement table CIDTBL in which the band sharing unit BAND is arranged.
M controls the WRR load based on the connection identifier arrangement table CIDTBL.

The common FIFO buffer system is described in, for example, Manabu Yoshino, Satoru Higashima, Norihito Ienaga "T
Examination of CP / IP Fairness ”, 99 IEICE B-8-1
5, in a scheme introduced in March 1999, a FIFO buffer common to all connections CONNCTN is provided, and a cell exceeding a threshold in the buffer is tagged, and the tagged cell is preferentially discarded. In both cases, tagging and discarding are performed on a frame basis (EPD: Early)
Packet Discard).

In this embodiment, the band sharing unit BAND
In M, it is assumed that the WRR method is used, but the following technology can be similarly applied to other methods. Here, the CID arrangement based on the connection identifier arrangement table CIDTBL in the band sharing unit BANDM is described as CID.
Call it a placement problem.

In the allocation of the group GROUP to the 1.2 Gb / s highway HYWAY, bandwidth sharing is not performed. However, in order to assign each group GROUP to the highway HYWAY within the range of the allowable CDV, a device is required, and this is also a technical problem. For group assignment, the multiplexing unit MUX refers to a group identifier arrangement table GIDTBL in which identifiers (GID) of each group GROUP are arranged. Here, GID
Call it a placement problem. The invention of the present embodiment provides an efficient solution of the GID placement problem.

FIG. 4 shows the connection CONNECT.
FIG. 3 is a conceptual diagram for explaining a hierarchical structure of N, group GROUP, and highway HYWAY, and there is a portion different from an actual device configuration. In FIG. 4, the band sharing unit BANDM
And the multiplexing unit MUX are drawn completely separately, and it seems that after allocating connections of all groups by the band sharing unit BANDM, each group is multiplexed by the multiplexing unit MUX. BANDM and multiplexing unit M
UX is fused. In this case, referring to the group identifier arrangement table GIDTBL, only the assigned group is connected to the connection identifier arrangement table CIDTBL.
WRR is performed according to the load.

Hereinafter, the contents of this embodiment will be specifically described. The configuration is as follows. First, GI
After explaining the D arrangement problem, an outline of the procedure of the GID arrangement will be described. Next, details of the GID arrangement procedure will be described. As a general outline, first, a method of calculating an ideal position in the GID arrangement will be described, and then, a treatment in a case where the ideal position cannot be arranged will be described. Furthermore, technologies relating to CDV and technologies relating to random numbers will be described as element technologies used in the GID arrangement.

Hereinafter, the GID arrangement problem in this embodiment will be described in detail with reference to FIGS. 1 to 3 and FIG. FIG. 5 shows an example of an input pattern of the GID placement problem. In the figure, TEST1 is a test pattern 1, GPRIO is a group priority, GPCR is a group band, and GPCRD is a group band (decimal expression). GPRIO and GPCR are also shown in FIG.

In the GID arrangement problem, first, the group GR
Band GPCRD and priority GPRIO for each OUP are given. These are the values that characterize the group as attributes of the group. In addition, a group identifier GID is assigned to each group as an identifier, but this is not assigned at the stage of input data.

The group band is represented by a decimal expression GPCRD or an integer expression GPCR. Group bandwidth is 1.198
Indicates the band of the group in the Gb / s highway HYWAY, and takes a value from 0 to 599.0 Mb / s in decimal representation. Here, the unit band at the time of band arrangement is 0.1 Mb / s
It is. That is, one slot S shown in FIGS.
Since LOT is 0.1 Mb / s, the group bandwidth is GPCR = GPPCRD (Mb / s) /0.1 (Mb / s
s) = GPCRD × 10. This group band GPC
R indicates the number of elements in the group.

The priority of a group is an index indicating the tolerance of deviation (CDV) from the ideal position when arranging the elements of the group. The higher the priority, the smaller the tolerance of deviation.
FIG. 6 shows an example of the priority of the arrangement group. The priority is high priority (High Priority), low priority (L
ow Priority, no priority (Non Prio)
and HP, LP, and NP, respectively.

The specification of the GID arrangement can be specified for each priority. For example, all high priority groups are arranged within CDV ≦ 7, and 80% of all groups including high priority are arranged with CDV ≦ 31.

The basic policy at the time of group arrangement is high priority.
Groups are arranged in the order of low priority to no priority, starting with the group with the largest bandwidth. However, if you try to place a low-priority large band group after placing all high-priority small band groups, you may not be able to place it.
The arrangement order is changed so that the low-priority large band group is arranged before the high-priority small band group.

In the pattern example TEST1 shown in FIG. 5, the priority GPRIO is high priority (HP), and the band GPPCRD is 1
There are eight groups of 6.7, three groups of priority GPRIO with low priority (LP), and a band GPCRD of 8.7. The problem of the GID arrangement problem is to arrange the group elements of each group evenly on the 1.2G highway HYWAY.

That is, the group elements are evenly arranged by writing the GID of the group in the GID arrangement table GIDTBL shown in FIG. 1 or FIG. In the GID arrangement table of FIG. 1, the GID is described as 1 for every 10 slots from the slot SLOT having the address ADR of 0. This means that the arrangement position of the element of the group whose identifier is 1 is arranged in the slot SLOT having the address ADR of 0, 10, 20,. Similarly, the elements of the group whose identifier is 2 are arranged every 13 slots from the position where the address ADR is 5.

Hereinafter, a basic policy of the GID arrangement in this embodiment will be described.・ Groups with the same band are arranged together. This is called grouping. -Arrange from the group of the group with the large band. -The same band group grouped is all slots (11980)
Are arranged on the basis of a base offset randomly selected from. The start offset in the group is all slots (11980)
), But when the ideal position is already arranged, a random selection is made from all the GIDs. GIDs in one group are all slots (1198
0). The optimum position is selected in consideration of the ideal position and the allowable CDV value. When the GID within the group cannot be arranged within the range of the allowable CDV value, the already set GID in the group is deleted, and the start offset is set to all slots (11980).
Randomly select from and reset. If the GID within the group cannot be set even if the start offset for the allowable number of times is changed, change the method of converting the GID interval into an integer and try again. If the GID within the group cannot be set even if the method of converting the allowable number of GID intervals into integers is changed, the allowable CDV value is changed.

The major feature of the GID arrangement algorithm in the present embodiment is as follows. -Random selection of arrangement positions by random numbers-Ideal arrangement positions within the same band group are equally distributed as much as possible-When the allowable CDV value is 7, the optimal arrangement position is selected by pattern matching near the ideal position-Adjacent to the same group GID Placement Prohibition Function Details of the GID placement algorithm in the present embodiment will be described below, focusing on these features.

Next, the procedure of the GID arrangement will be described.
First, the following test pattern of TEST2 is taken as an example of a group hierarchy. ([HP] 167, 167, 167, 167, 167,
167,167,167,163,163,163,1
63,163,163,163,163,157,15
7,157,157,157,157,157,15
7,151,151,151,151,151,15
1,151,151,149,149,149,14
9,149,149,149,149,139,13
9,139,139,139,139,139,13
9,137,137,137,137,137,13
7,137,137,131,131,131,13
1,131,131,131,131,127,12
7,127,127,127,127,127,12
7, 113, 113, 113, 113, 113, 11
3,113,113,61,61,61,61,61,
61, 61, 61, 2, 2, 2, 2, 2, 2, 2, 2, 2)

This is a pattern composed of 12 groups of 96 groups of 8 pieces each, and is hierarchized as shown in FIG. A group of the same band group is referred to as a same band group group. At the time of group arrangement, the arrangement processing is performed for each group from the group with the largest band, and the arrangement is performed for each group in each group, and for each individual GID in each group. As the groups are hierarchized, the arrangement processing of the groups is also hierarchized. FIG. 8 shows the basic configuration of the GID arrangement, and FIGS. 9 to 12 show the procedure of processing from trunc_s0 to trunc_s3 in the figure.

Next, the calculation of the ideal position of the GID arrangement in the GID arrangement of the present embodiment will be described. First, equal distribution of ideal arrangement positions in the same band group will be described. The basic principle of the GID arrangement algorithm in the present embodiment is to distribute the ideal arrangement positions of the elements of the grouped group having the same band as evenly as possible among all the GIDs. Hereinafter, a method of the equal distribution will be described.

For example, the test pattern TES shown in FIG.
The band GPCR of the first eight data of T is 167. These eight are collectively processed as a group (grouping).
Here, 167 group elements are assigned to all slots 11
To allocate as evenly as possible during 980, 11980 /
A group may be arranged every 167 = 71.7365 slots, but one slot cannot be divided and the GID address is an integer value, so the group arrangement must be performed under this restriction. Here, the allowable CDV value indicates how much any of the strict ideal calculation positions is allowed. The allowable CDV value and the CDV calculation method will be described later in detail.

Next, the arrangement of the steps will be described. Now, to arrange 167 groups in 11980 GIDs in an integer manner, 11980 = 71 × 44 + 7
From 2.times.123, it is necessary to arrange the 44 at 71 intervals and the 123 at 72 intervals in a mixed manner. For convenience, the smaller interval (71 in this case) is called step1, and the larger interval (72 in this case) is called step2. s
step2 = step1 + 1.

In this embodiment, step 1 and step 1
The calculation of p2 is performed by the function cnt_mix_clc_int. FIG. 13 shows the cnt_mix_clc_int
6 is a flowchart showing the procedure of FIG. Function cnt_mi
x_clc_int is the process trunc_s1 in FIG.
_S1 cnt_set_step_inf and cn
Used in t_set_size_inf.

Cnt_mix_clc_int in FIG.
Is a process of allocating as many divided_int_elements as possible within the whole number whol_int to be divided, as much as possible, only in the world of integers. Intervals between placement elements size1 and size2 = size
1 + 1, number num of size1, num of size2
Returns m2. In this example, whole_int = 1
1980, divider_int = 167, size
1 = step1 = 71, size2 = step2 = 72
Becomes

Processing cnt_mix_clc_int in FIG.
In _s1, size1 and size2 are obtained, and processing c
In nt_mix_clc_int_s2, whole
The remainder of _int / divider_int, 123 in this case, is obtained. Processing cnt_mix_clc
In _int_s3, the number num1 of size1 is 44 in this case, and the number num2 of size2 is 123 in this case.

As a method of mixing the two types of intervals of step 1 and step 2, the following method can be considered. 1) First, step1 is arranged collectively, and after step1 is arranged, step2 is arranged. 2) The mixing ratio of step 1 and step 2 is calculated, and step 1 and step 2 are mixed and arranged according to the calculated ratio. Since the mixture ratio is an integer, the remainder is left at the end, and either one is solidified and arranged. 3) The step cumulative value (GID address) is calculated as a real number by the following equation (1) without considering step1 and step2, and the real value is rounded to an integer value. GID address = 11980 / band × i (1) where i = 1, 2,... Band The algorithm is complicated in the order of the methods 1) to 3), but the deviation from the ideal GID value is small.

In the GID arrangement algorithm according to this embodiment, all the methods are used in the order of 1) to 3). However, in order to suppress the CDV by integer conversion to less than 1, 3 methods are finally used. ) Method is used. Although the method of 3) seems simple, it is not so easy under the constraint that the variables used are limited to positive integers as much as possible.

FIG. 14 shows a simulation result in which the maximum value, the minimum value, and the average of the CDV and the borrow are obtained when the GID address converted into an integer using each method is obtained when the group bandwidth is 380. is there. Band = 380 means that 380 group elements are arranged as equally as possible from GID 0 to 11979,
The table shows the maximum / minimum / average of the CDV and the borrow value in the 380 arrangements.

In the method 1), first, step1 = 31 is set to 1
After arranging all 80, next step2 = 32 is set to 20
0 (11980 = 31 × 180 + 32 × 20)
0). Since step 1 with the smaller interval is arranged first, no borrowing occurs, but the CDV becomes very large, which is 100 times that in the case of method 3).

In the method 2), 380/180 = 2.11
Step 1 and s are based on the integer ratio 2 which rounds 11
Steps 2 are alternately mixed one by one, and finally the remaining 20 steps 2 are arranged together. 0.47 of borrowed
Reference numeral 37 denotes a borrow that occurred because step 2 instead of step 1 was initially allocated, but remained until the end, and no new borrow has occurred thereafter. Also in the case of the method 2), the CDV is large, which is ten times that of the method 1).

In method 3), four methods of rounding, rounding down, rounding up, and randomly selecting rounding up and down were tried as methods of converting real numbers to integers. If the same method is used consistently for rounding, truncation, and rounding, C
It can be seen that DV is almost the same and less than 1. Borrow,
The value is zero when rounding down to always select a smaller integer value, a value close to 1 when rounding up to always select a larger integer value, and intermediate when rounding. Also, in the method in which the round-off and the round-up are selected at random, the CDV becomes 1 larger than the other methods, so that it is understood that the same method must be used consistently for rounding of the integer.

In the present embodiment, these functions are a real number mixed parameter calculation setting processing function cnt_mix_
Realized by real. FIG. 15 shows the function cnt_mix
It is a flowchart which shows the procedure of _real. Function c
nt_mix_real is trunc_s1 in FIG.
_S1 cnt_set_size_inf and tr
Used in unc_s1_s4. Function cnt_mix
_Real is the whole number whole_i from which the division is based
In the nt slot, a process of arranging the elements of the division number divider_int as evenly as possible is performed in the real world, and an integer value is written to the arrangement table and returned. Input and output from the outside are integers, and real numbers are closed only in this function. It has a function to check whether the number of arrangement intervals matches the case of integer calculation.

Processing of FIG. 15 cnt_mix_real_
In s1, the integer variables whole_int and divide
r_int is converted to a real number. Processing cnt_mix
In _real_s2, a rounding bias is calculated.
The rounding bias will be described later. Processing cnt_mix
_Real_s4, i-th real number arrangement position answ
er, and rounding bias round_bia
s is added, rounded down to an integer, and answer_int
Is obtained and set in the table * mix_cum_tbl. Processing cnt_mix_real_s3 to cnt
In the loop of _mix_real_s7, i is changed from 1 to di
It is changed up to vider_int. Processing cnt_
In the mix_real_s5, the interval and the number obtained here are the integer calculation function cnt_mix_clc in FIG.
It is checked whether the value matches the value obtained by _int.

Next, the change of the step mixing method will be described. First, the addition of the rounding bias will be described. Based on the formula (1) of the above method 3), the addition value of the step when the band = 380 is calculated as a real number, rounded off, truncated,
Fig. 1 shows the difference in the integer value when the integer is rounded up in three ways.
Table 6 shows. As can be seen from FIG. 16, the GID address, that is, step 1 and step
The mixed state of No. 2 is slightly different.

In the algorithm of the present embodiment, when all the GIDs cannot be arranged even if various start offsets are tried at the time of arranging the individual GIDs of the group, further arrangement is made by utilizing the difference of the step mixing method. There is a function to try. This is realized by a variable in the mixed parameter calculation setting processing function cnt_mix_real by a real number and a rounding bias round_bias in FIG.

The rounding bias round_bias is obtained by round_bi in the processing cnt_mix_real_s2.
Determined by as_int / bias_pitch. This bias_pitch is 100, round
Since _bias_int takes an integer between 0 and 99, the rounding bias round_bias is between 0 and 0.9.
Numbers in increments of 0.01 up to 9 can be taken.

Here, at the time of integer conversion, a rounding bias round_bias is added to the original real value, and then rounded down, thereby giving a variation of integer conversion. Rounding bias round_bias = 0 rounds off and 0.5 rounds off. As shown in FIG.
Rounding up cannot be represented by rounding bias addition because the handling of end points is different, but rounding bias round_bias
It is simulated by setting to 0.99.

The rounding bias is obtained from the ROM table o_cnt
Round_bi described in _round_seq
It is set according to the value of as_int. This is as follows, but only the first five are actually used.

{0,99,50,25,75,12,3
7, 62, 87, 6, 18, 31, 43, 56, 68,
81,93,3,9,15,21,28,34,40,
46, 53, 59, 65, 71, 78, 84, 90, 9
6,2,5,8,11,14,17,20,23,2
7, 30, 33, 36, 39, 42, 45, 48, 5
2,55,58,61,64,67,70,73,7
7,80,83,86,89,92,95,98,1,
4, 7, 10, 13, 16, 19, 22, 26, 29,
32, 35, 38, 41, 44, 47, 51, 54, 5
7,60,63,66,69,72,76,79,8
2,85,88,91,94,97,24,49,7
4｝

There is such a step mixing function in order to obtain a variation of the step mixing. The step mixing function is very effective when the steps are initially arranged by the method 1), but relatively ineffective when the steps are arranged using real numbers by the method 3).

Next, the matrix coordinates will be described. If step 1 and step 2 for the number of bands are appropriately mixed by any method, it is necessary to store the result of this mixing. It is troublesome to memorize the result of mixing, so if you remember only the basic rules of step mixing, it seems that the result can be reproduced immediately without memorizing it, but it is not so easy. You only need one band group,
When a plurality of groups exist and are arranged as a group, the arrangement start positions are different for each group.
If the mixing method of ep2 is not performed in the same framework, a phenomenon occurs in which a slight shift is accumulated and a cluster of the arranged GID groups is formed in a part. In order to solve such a problem, the embodiment introduces the concept of matrix coordinates.

The matrix coordinates are defined for each band, and are used for the purpose of storing a method of mixing step 1 and step 2. Matrix coordinates (mtx
Based on (i, mtxj) = (0, 0), the mtxj-th GID of the mtxi-th step is (mtxi, mt
xj).

FIG. 18 shows a conceptual example of matrix coordinates when the band = 167. In the figure, GIDTBL represents a group identifier arrangement table. In FIG. 1, GIDTBL drawn in one horizontal row is divided into lengths of step1 or step2, and drawn in a column. The ORG in the figure indicates the origin at this time, and the coordinates in the row direction of the matrix coordinates are mtx.
i, the coordinates in the column direction of the matrix coordinates are represented by mtxj. It should be noted here that the origin ORG of the matrix coordinates is not necessarily a slot having a GID address of 0.

FIGS. 19, 20, and 21 show correspondences between the matrix coordinates (mtxi, mtxj) and the GIDs from 0 to 11979. FIG.
Is the origin of the matrix coordinates (mtxi, mtxj) = (0,
0) corresponds to GID = 0. Since this serves as a reference for the matrix coordinates, this case is first calculated when setting the matrix coordinates. Although the matrix coordinates are stored in association with the GID address, the matrix coordinates (mtxi, mtx
j), the GID address is obtained from (mtx
i, mtxj) does not need to be stored for all combinations of mtxi when mtxj = 0.
It is only necessary to store (mtxi, 0) which is D. In FIG. 19, only the value of the shaded portion is a pointer * (a
_Cnt_step_cum_tbl + mtxi).

FIG. 19 is a reference for setting the matrix coordinates. Actually, the origin of the matrix coordinates (mtxi, mtxj) = (0,
0) does not correspond to the GID address = 0, but corresponds to a base offset address selected from 0 to 11979 with a random number each time. FIG. 20 shows an example in which the origin of the matrix coordinates is set to the base offset address = 111727 in the example of FIG. FIG.
Is the base offset = 1172 in each coordinate value in FIG.
It is obtained by adding 7. In the algorithm,
Function cnt_matrix_coordi shown in FIG.
The table of FIG. 20 is used instead of FIG.

Function cnt_matrix_coordi
Nate_set is trunc_s1_s4 in FIG.
Used in FIG. 19 is necessary in the process of calculating FIG. 20, but need not be stored once FIG. 20 is calculated. In FIG. 20, similarly to FIG. 19, the GID of each mtxi when mtxj = 0 (mtxi, 0), that is, only the shaded value is a pointer * (a_cnt_step_a).
dr_tbl + mtxi).

By the way, it is understood that FIG. 20 should be used when obtaining the corresponding GID from the matrix coordinates (mtxi, mtxj). Conversely, when obtaining the corresponding matrix coordinates (mtxi, mtxj) from an arbitrary GID, FIG.
* (A_cnt_step_adr_tbl + mt of 0)
The difference can be obtained by searching xi), but it is convenient if there is a correspondence table for obtaining the matrix coordinates from the GID. This is shown in FIG. In FIG. 21, GID = 0 is (mtx
i, mtxj) = (3, 38), and (mtxi,
(mtxj) = (0,0) corresponds to GID = 111727. The matrix coordinates (mtxi, mtxj) in FIG.
a_cnt_gid_tbl [i]. mtxi and a_c
nt_gid_tbl [i]. mtxj.

The method of mixing the steps when one group of a certain band is arranged has been described above. Next, an arrangement method when a plurality of groups of the same band are arranged will be described. If there are a plurality of groups of the same band, they are arranged separately and sequentially instead of being arranged separately.

FIG. 23 shows an example of an ideal position in the case where one group of the band 167 is arranged, and FIG. 24 shows an example of an arrangement in which two groups are arranged. In FIG. 23, step 1 and step 2 are mixed and arranged as described above. At this time, the GID address of the origin of the matrix coordinates is a base offset selected by a random number, but in FIG. 23, the base offset is set to 0 for simplicity.

In FIG. 24, group 1 and group 2
Group is arranged, but here Group 1 and Group 2
In order to disperse as much as possible, as shown in the figure, it is better to arrange the group 2 near the middle of the steps of the group 1. Further, when the number of groups increases, if the groups are arranged so as to be evenly distributed in the steps, the possibility of collision decreases. In the same band group group, the base offset is made common and arranged based on the same matrix coordinates. This means that using matrix coordinates, mtxj
Can be realized by increasing the value of mtxi by one while keeping the constant. This is shown in FIG. There are two problems here. 1) Selection of mtxj corresponding to a plurality of groups 2) Selection of mtxi to start arrangement These will be described below.

Next, the concept of size will be described. For mtxj corresponding to a plurality of groups, basically, step1 is equally distributed and allocated. In the present embodiment, 11980
Step of dividing the GIDs into bands.
An interval obtained by dividing a step by the number of the same band group is called a size. Just as the value of the step is step1 or step2 = step1 + 1, the size is also size1 or size2 = size1 + 1.

The step and the size are similar in that the divided matrix is divided into integers, and can be compared as follows. By utilizing this similarity, in the present embodiment, the function cnt_mix of FIG.
_Clc_int and the function cnt_mix_ in FIG.
Using real, the step information is cnt_set_s
In step_inf, the size information is stored in cnt_set_si
It is calculated with ze_inf. Function cnt_set_s
FIG. 26 shows the procedure of step_inf, and the function cnt_set
FIG. 27 shows the procedure of _size_inf. These functions are used in the process trunc_s1_s1 of FIG.

The step information is represented by * (a_cnt_step)
_Cum_tbl + i), the size information is * (a_cnt_size_cum_tbl +
i). As shown in TEST2 of FIG.
FIG. 29 shows an example of the size information holding table when there are eight groups of 167, and FIG. 30 shows the arrangement of the example.

The size information indicates the arrangement within one step, and indicates the matrix coordinates mtxj common to all steps. Here, the size information is smaller step 1
(71 in the example of FIG. 30), the last GID in the case where the step is step 2 (i.e., i in the example of FIG. 30).
Mtxj at the right end of the step in the case of = 1,2,166 =
In 71), the group arrangement as an ideal position is not performed.

Next, the uniform distribution of the same band group start step will be described. In FIG. 30, i = 0, 1, 2,..., 16 for mtxj assigned to each group.
6, 167 group elements are arranged. However, the arrangement position CALP shown in black in FIG. 30 is only an ideal position that it is desirable to be able to arrange there. If the ideal position CALP has already been arranged, it is necessary to search for empty positions before and after and locate the empty position within an allowable CDV range.

When assigning GIDs of a large number of groups, if the first assignment, the ideal position is not already arranged, but if it is the second or later, the ideal position may already be arranged. Sex comes out. In view of this possibility, the step of starting group arrangement, that is, FIG.
It can be understood that which mtxi column to start with in GIDTBL of 0 also has significance. Referring to FIG. 30, GI starts from the step of mtxi = 0 for all groups.
Starting the D placement is the simplest. At the beginning of the arrangement, the number of existing slots is small and the arrangement at the ideal position is possible. However, as the arrangement progresses and mtxi increases, the number of the already arranged slots increases and the arrangement at the ideal position becomes difficult.

Since the same occurs in all the groups, the number of empty slots decreases as the step of mtxi = 166 is approached, and in some cases, it may be impossible to arrange within the allowable CDV. Here, in order to avoid such a situation, the arrangement start positions are shifted for each group so as to be distributed as much as possible in all the GIDs. The bias number bias_num represents this deviation.
It is obtained by the following equation. However, the fraction is rounded down to an integer value. Bias number bias_num = band / number of same band groups In the example of FIG. 30, the number of biases = 167/8 = 20. Therefore, the arrangement start step of each group is i = 0,2
0, 40, 60, 80, 100, 120, 140. This state is shown in FIG.

A reference point at the time of group arrangement is called an offset. Each offset will be described with reference to FIG. First, the base offset base_offset sets the origin ORG of the matrix coordinates.
A random number is selected from the D addresses (0 to 11979).

The arrangement start position of each group is a start offset start_offset. To set the start offset, first, a start offset origin start_offset_org is set for each size in which the first step of mtxi = 0 is equally distributed with the same band group number based on the base offset. Step i is sequentially shifted from the start offset origin of each group by the number of biases bias_num to set a start offset start_offset.

Next, the start offset setting function cnt
_Set_start_offset will be described.
As described above, the ideal arrangement positions are equally distributed, and matrix coordinate conversion tables are stored in FIGS. 19, 20, and 21. At the time of GID arrangement, these are read and used.

The start offset setting function used in the start offset arrangement is cnt_set_star.
t_offset. FIG. 32 shows a function cnt_se
4 shows a flowchart of t_start_offset. This function corresponds to the process trunc_s2_s1 in FIG.
Is used for setting the start offset. Here, the point is that the variable loop2 indicates the number of the group in the group.

The k of the start offset org shown in FIG.
Is the k-th value * (a_cnt_size_cum_t) of the size information holding table of FIG.
bl + k). In addition, * in FIG.
Using (a_cnt_step_cum_tbl),
The start offset k is obtained by adding the number of biases bias_num × loop2. This is indicated by * in FIG.
Bias of (a_cnt_step_adr_tbl)
_Num × loop second value and the size information holding table in FIG. 29 * (a_cnt_size_cum_tb)
It can also be obtained by using the second value of loop in 1).

Next, arrangement GID acquisition processing cnt_set
_Clc_gid will be described. It is the arrangement GI that reads out the equal distribution information when the individual GIDs are arranged in Loop3.
D acquisition processing function cnt_set_clc_gid. Start offset and loop3 (number of individual G
ID is set), an ideal arrangement position is obtained as shown in FIG. This function corresponds to the process “run” in FIG.
Used for obtaining the ideal position of c_s3_s1.

First, the processing cnt_set_clc_ shown in FIG.
The mtxi and mtxj coordinates of the start offset are obtained from the table a_cnt_gid_tbl for obtaining the matrix coordinates from the GID in FIG. 21 using gid_s1. Next, in processing cnt_set_clc_gid_s2, loop3 is added to the mtxi coordinates to obtain the mtxi coordinates of the GID to be currently arranged.

Next, processing cnt_set_clc_gi
In d_s3, the mtx of the GID currently being arranged
The j coordinate is set equal to the mtxj coordinate of the start offset. Further processing cnt_set_clc_gid_
At s4, the mtxi coordinate and the mtxj coordinate of the GID to be currently arranged are obtained, so the GID address of the slot is obtained using the step information table a_cnt_step_adr_tbl in FIG.

Next, the same group GID adjacent arrangement prohibition function will be described. The algorithm according to the present embodiment has a restriction that GIDs in the same group must not be arranged adjacently. In a group with a large band, even if a GID is arranged within the allowable CDV range,
A phenomenon occurs that they are arranged adjacent to each other. It is preferable that there is no adjacent arrangement because GIDs within a group are dispersed.

The determination of the prohibition of the adjacent arrangement is performed by determining whether the GID can be arranged (prohibition of the adjacent GID) function cnt_chk_gid_
It is made available. In FIG. 34, cnt
13 shows a flowchart of _chk_gid_available. This corresponds to the process trunc_s3_s in FIG.
2 or trunc_s3_s4 or the like is used to determine whether individual GIDs can be placed in loop3.

Hereinafter, a description will be given of a process when a group element cannot be arranged in a slot at an ideal position. In the previous section, the mechanism of equal distribution of ideal positions was described. However, the ideal position CALP is a position that is "desirable to be able to be arranged there" to the last, and if that slot has already been arranged, another slot must be searched for. FIG. 36 is a diagram illustrating an ideal position and its periphery. GID arrangement table GIDTB of FIG. 36
L indicates a part of GIDTBL in FIG.
In the figure, if it is assumed that a new GID cannot be arranged because the slot at the ideal position CALP has already been arranged, it is necessary to arrange the GID in an appropriate slot in the vicinity.

The method of searching for another arrangement slot differs depending on the type of object to be arranged. Even if the target ideal position is vacant, it may be better to arrange the GID around the target GID instead of at the ideal position in consideration of the surrounding GID arrangement situation. These distinctions are described below.

FIG. 35 shows a summary of measures to be taken when a slot cannot be placed in an ideal position slot.

Next, pattern matching when CDV = 7 will be described. In the algorithm of the present embodiment, when the allowable CDV value is 7, the optimal arrangement position is selected by pattern matching near the ideal position. That is, FIG.
From trunc_s3_s2 to trunc_s3_
If the allowable CDV value is 7 instead of s5, pattern matching is performed in the following procedure.

The allowable CDV value of 7 means that if it can be arranged in eight peripheral slots including the ideally located slot, it is acceptable. If each GID address is empty and 0 if it is arranged, the allowable CDV
The range with a value of 7 can be represented by an 8-bit bit pattern.

Further, it is desirable to consider the situation before and after this range, but it becomes complicated if an extremely wide range is considered. Therefore, in this algorithm, one address is added before and after each range, and the range of the 10-bit pattern is considered. I have. By expressing the degree of vacancy of the ten GID addresses around the ideal position with a 10-bit bit pattern, the optimal arrangement position according to the degree of vacancy is selected in correspondence with the 10-bit bit pattern. An overview. The 10 bit empty condition is 1 even if all patterns are covered.
Because of the 024 pattern, the optimum arrangement position can be manually selected.

In the initial algorithm, if the ideal position is empty, the pattern is arranged unconditionally, and the allowable CDV value does not consider each bit before and after the range of 7.
There were 128 bits. However, a phenomenon in which the arranged GIDs are continuous due to the arrangement at the ideal position, and a disadvantage of increasing the number of arranged GIDs outside the range under consideration occur, so that the 10-bit pattern matching is performed. Was.

The policy for selecting the optimum arrangement position from the already arranged patterns is as follows. 1) When there are a plurality of vacant positions (bit 0) that are equally suitable, all of them are set as arrangement candidates, and one of the candidate positions is selected by random numbers during pattern matching. 2) If there is a vacant position (0) where 1 (placed) continuation does not occur due to the new placement, place it there. In other words, when there are 0 (empty) continuations of 3 bits or more, they are arranged at 0 positions excluding the 0s at both ends. 3) If there is no empty position (0) where 1 (placed) continuation does not occur due to the new placement, select a position where the number of 1 (placed) continuations is the minimum. 4) An arrangement position is selected by making pessimistic assumptions about the arrangement state outside the 10-bit range. In other words, depending on the arrangement status on the outside, even if one continuous number smaller than one continuous number within 10 bits can be realized, if the outside is another situation, a longer continuous one may occur. Do not select a position.

The ROM table o_cnt_cdv7_pa
FIG. 37 shows t and a part of the bit pattern. The left half of the table is a ROM table, and the right half is a 10-bit arrangement pattern. In the ROM table, one row of data is a set, and represents the optimal arrangement position of the arrangement pattern on the right side. The pattern code in the leftmost column of the ROM table indicates a number when the arrangement bit pattern is regarded as a binary sequence and converted to a decimal number.

In the first row of FIG. 37, since the bit pattern is (0,0,0,0,0,0,0,0,0,0),
The pattern code is 0. Pattern code 25
6 is (0,1,0,0,0,0,0,0,
0,0), and the pattern of the pattern code 1023 is (1,1,1,1,1,1,1,1,1,1).

The weight of each bit at the time of creating the pattern code is shown at the top of the arrangement pattern table in the right half of FIG. The label of each bit is described at the top of the arrangement pattern table. The bits of the bit pattern are not named 0, 1, 2,... From the right side, but are numbered alternately left and right from the ideal position in the middle. This indicates the order when starting from the ideal position and sequentially searching for a position close to the ideal.

In the ROM table, the optimum position when a new 1 is arranged in the current bit pattern is shown. When there are a plurality of suitable positions, labels of all candidates are listed. . The number of candidates is described in the second column of the ROM table. The pattern code 0 in the first row indicates that eight candidates of candidate labels 0, 1, 2, 3, 4, 5, 6, and 7 are similarly suitable. The positions of the labels listed as the arrangement candidates in the ROM table are indicated by oblique lines in the arrangement pattern. If you have multiple placement options,
One is selected by a random number at the time of actual arrangement.

When the pattern code is 0 and there are many empty bits, there are many arrangement candidate positions (shaded portions), but as the pattern codes 1, 2, 3,... Increase, the arrangement candidate positions decrease. . In a pattern having many 1s as in the pattern code 255, the arrangement candidate is 1 in the label 6.
Only bits, but like pattern code 256,
In a pattern with many 0s, the arrangement candidate position has 6 bits. Since all bits are 1 in the pattern codes 1022 and 1023, the arrangement candidate is zero and cannot be arranged.

In addition to these functions, the one in which control such as arranging only at odd addresses in the initial stage of arrangement is added.
CDV allowable range pattern matching function cnt_chk_c
This is realized by dv_pat_value. The one in which a variation is added with more detailed control added to the arrangement address is cnt_chk_cdv_pat.
_R1, cnt_chk_cdv_pat_R13, c
nt_chk_cdv_pat_R135, cnt_c
hk_cdv_pat_R1357.

Function cnt_chk_cdv_pat_v
FIG. 38 is a flowchart showing the procedure of “alue”, and FIG. 39 is a flowchart showing the procedure of the function cnt_chk_cdv_pat_R1357. Function cnt_ch
k_cdv_pat_R1 to cnt_chk_cdv
_Pat_R135 is a function cnt_chk_cdv_
Since it is a subset of pat_R1357, the procedure is included in FIG.

FIG. 38, cnt_chk_cdv_pat
_Value is the GID of the grant when CDV ≦ 7
Is read, and the arrangement status of the surrounding 10 slots is read and expressed in a 10-bit pattern ind_cdv7. This pattern is collated with a pattern table (stored in a ROM table) in which the optimum position has been searched in advance, and is arranged at the optimum position.

Processing cnt_chk_cdv_p of FIG. 38
at_value_s6 and cnt_chk_cdv_p
at_value_s7 and cnt_chk_cdv_p
In at_value_s8, the rom table o_cnt_cdv7_pa
t, o_cnt_cdv7_pat_o, and o_cn
Although t_cdv7_pat_e is selectively used, each is a content table as shown in FIG.

Function cnt_chk_cdv_p in FIG. 39
from at_R1 to cnt_chk_cdv_pat_R1
In 357, when the total set bandwidth is small, the group element is arranged at the GID of the odd address.

Function cnt_chk_cdv_pat_R
1, the GID address is arranged at the position of the remainder 1 when divided by 8 (processing cnt_chk_cdv_pa in the figure)
From t_R1357_s11 to cnt_chk_cdv_
pat_R1357_e1). If placement is not possible even after trying the allowable number of times, the process proceeds to a function to be placed at the next remainder 1 or 3.

Function cnt_chk_cdv_pat_R
In 13, the remainder when the GID address is divided by 8 is 1
Alternatively, it is arranged at the position 3 (processing cnt_chk in the figure)
_Cdv_pat_R1357_s31 to cnt_c
hk_cdv_pat_R1357_e3). If placement is not possible even after trying the allowable number of times, the process proceeds to a function to be placed at the next remainder 1 or 3 or 5.

Function cnt_chk_cdv_pat_R
At 135, the remainder when the GID address is divided by 8 is placed at the position of 1, 3, or 5 (processing cnt in the figure).
_Chk_cdv_pat_R1357_s51 to c
nt_chk_cdv_pat_R1357_e5).
If placement is not possible even after trying the allowable number of times, the process proceeds to a function to be placed at the next remainder 1 or 3 or 5 or 7.

Function cnt_chk_cdv_pat_R
In 1357, the remainder when the GID address is divided by 8 is placed at the position of 1 or 3 or 5 or 7 (processing cnt_chk_cdv_pat_R1357_s in the figure)
71 to cnt_chk_cdv_pat_R1357
_E7). If placement is not possible even after the allowable number of attempts, give up the odd address placement.

[0117] Other trial procedures when it is not possible to arrange in the ideal slot include the following. The random numbers will be described later in detail. If the start slot of trunc_s2_s1 in FIG. 11 cannot be located in the ideal slot, the GID random number search process (all GIDs) shown in FIG.
Target) function cnt_chk_start_rand_a
11 is used. This function converts a random number with a period of 16384 into a random number between 0 and 11979 as shown in the expression (3) described below, and a maximum of 1638 until an empty GID is found.
Try 4 times. If there is even one empty GID slot, it is always found.

If the base slot cannot be placed in the ideal slot due to the setting of the base offset of stunc_s1_s2 in FIG. 10, the function GID random number search processing (for all GIDs) ± del
ta, cnt_chk_start_rand_all
_Delta is used. This function corresponds to cn shown in FIG.
Similar to the t_chk_start_rand_all function, an empty GID is searched using a random number, but processing cnt_chk_
According to the determination of star_rand_all_s4, the GI
An empty GID is searched not only for D but also within a range of ± delta.

In the process trunc_s3_s3 of FIG. 12, if the allowable CDV value exceeds 7, and if it cannot be arranged in the ideal GID slot, the random number GID search process cnt_chk_cdv_ in the CDV allowable range shown in FIG.
rand is used. In this function, a GID within the allowable range of the CDV is searched by a random number.

If the group elements cannot be arranged within the allowable CDV value even after repeating the processes trunc_s3_s3 to trunc_s3_s5 in FIG.
The process proceeds from c_s3_s6 to trunc_s3_s61 to reset the group element of the group that has been arranged, and from the process trunc_s3_e6 to trunc in FIG.
Returning to c_s2_st2, the start offset is changed and the group element is reset.

If all the elements of the groove cannot be set even after performing the allowable number of attempts to change the start offset by returning from the loop 3 to the loop 2, the processing t of FIG.
The process proceeds from trunc_s3_s7 to trunc_s3_s71, and from the process trunc_s3_e7 to tr in FIG.
Returning to unc_s1_st2, the rounding bias is changed and the group element is reset. The change of the rounding bias has been described above.

If all the elements of the groove cannot be set even after performing the allowable number of attempts to change the rounding bias by returning from loop 3 to loop 1, the processing of FIG.
The process proceeds from c_s3_s8 to trunc_s3_s81, changes the allowable CDV value, and performs processing trunc_s3_
Returning from e8 to trunc_s2_st2 in FIG. 11, the group element is reset. To arrange group elements without increasing the allowable CDV value as much as possible,
Trials from c_s3_s6 to trunc_s3_s7 are performed, but if still not possible, trunc_s3
The allowable CDV value is increased at _s81. Acceptable CD
The method of increasing the V value will be described later. Although the maximum value of the allowable CDV value is all slots 11979, if the slot cannot be arranged even here, trunc_s3_ed terminates as unplaceable.

Next, the CDV in the present embodiment will be described. The purpose of the CID placement algorithm of this embodiment is to distribute each element of the group as evenly as possible among all GID slots. But usually,
The number of all GID slots cannot be divided by the number of group elements. On the other hand, the GID address is an integer value. Therefore, the GID address of the ideal arrangement slot needs to be obtained by converting a strict ideal calculation position (real number) into an integer, and a slight shift occurs here. Although this shift always occurs due to the structure of the GID arrangement problem, it is necessary to grasp the mechanism and range of occurrence and keep it within a desired range. This shift is called structural CDV. The structural CDV will be described later.

Also, at the time of group element arrangement, the GID slot at the ideal arrangement position is not necessarily empty, and in this case, the GID slot is arranged around the ideal arrangement position. At this time, the range allowed as the periphery of the ideal arrangement position is the allowable CDV value. The allowable CDV value will also be described later.

Next, the mechanism of CDV and the structural CDV will be described. First, the structural CDV will be described. Here, the concept of borrowing (debt) is introduced for CDV calculation. Also, in order to distinguish it from the normal CDV calculation method, the CDV is referred to as a narrowing here. Narrowing refers to the difference between the ideal position and the actual position when the actual position precedes the ideal position. When the actual arrangement position is delayed from the ideal position, the narrowing becomes zero and it seems that there is no penalty, but a new borrow occurs. Here, the actual arrangement position is an alternative position that is actually arranged when it cannot be arranged at the ideal position. In addition, “preceding” means being arranged on the left side (small slot address) in the GID table, and “delaying” means being arranged on the right side (larger slot address) in the GID table.

That is, the ideal position of the next and subsequent arrangements is moved backward by the amount of the borrow, and if they are actually arranged at the calculated ideal positions, the narrowing will occur. Therefore,
The actual ideal position is a position obtained by adding (delaying) the sum of the borrowed data to the calculated ideal position. Here, the relationship between the ideal position and the actual position is referred to as an index. Depending on the sign of the indicator, narrowing or new borrowing will occur.

FIG. 42 is a flowchart of the CDV calculation algorithm cnt_clc_cdv_show according to the present embodiment. This function is used to calculate the CDV when the GID arrangement is performed.
nc_s3_s21 and trunc_s3_s41,
Processing cnt_chk_cdv_pat_val of FIG. 38
ue_s11, processing cnt_chk_cdv_ in FIG. 39
pat_R1347_s12, cnt_chk_cdv
_Pat_R1347_s32, cnt_chk_cd
v_pat_R1347_s52, cnt_chk_c
Used in dv_pat_R1347_s72.

In FIG. 42, processing cnt_clc_c
In dv_show_s1, index = (calculation ideal + borrow)
The index is calculated by:-actual position = borrow + (calculation ideal-actual position). Process the result cnt_clc_cdv_sh
ow_s2, and when the index is zero or positive, the ideal position is later than the actual position, so the process cnt_clc_c
The process proceeds to dv_show_s3, and narrowing is set as an index. At this time, no new borrowing occurs, and the process cnt_clc_
It is calculated by cdv_show_s5, and the borrowing integrated value, that is, Σ
Borrowing is immutable. If the index is negative, the ideal position is earlier than the actual position, so the process cnt_clc_cdv_sho
The process proceeds to w_s4, and narrowing is set to zero. At this time, the absolute value of the index is newly borrowed, and this value is processed by cnt_clc.
_Cdv_show_s5 is added to $ borrow.

Next, the mechanism of CDV calculation will be described with reference to FIG. 43 showing an example of CDV calculation. In the figure, BA
NDAX indicates the band axis. This corresponds to the group identifier arrangement table GIDTBL in FIG. The group identifier arrangement table GIDTBL in FIG.
In contrast to the digital one in units of
The band axis BANDAX of No. 3 is analog. GID arrangement is performed sequentially from the left side of the band axis. Other, CA
LP is the calculated ideal position, REALP is the actual position, SFTDP
Is the shifted ideal position, NEWDBT is a newly generated new borrow, SUMDBT is $ borrow, that is, the integrated value of the previous borrow, and CDV is narrow. The number added to each symbol indicates the order of each value set in the order of 0, 1, 2, 3,... Over time.

In FIG. 43, calculated ideal position CALP
0, CALP1, CALP2, and CALP3 are arranged at equal intervals. Since the real position REALP1 is earlier than the CALP1, the narrowing occurs here. That is, since index 1 = calculated ideal CALP1-real position REALP1 ≧ 0, CDV1 = index 1. At this point,
No borrowing has yet occurred.

The actual position REALP2 is the calculated ideal position CA
Since it is slower than LP2, a new borrow occurs here. That is, since index 2 = calculated ideal CALP2−real position REALP2 <0, new borrowing NEWDBT2 = −1 × index (positive value). Since the borrow has been made, the ideal position 3 becomes the shifted ideal position SFTDP3 obtained by adding Σborrow (only the new borrowing NEWDBT2 at this time) to the calculated ideal position CALP3.

In the case of the real position REALP3a, the index 3a
= Shifted ideal position SFTDP3-real position REAL
P3a ≧ 0, and CDV3a = index 3a. In the case of the actual position 3b, the index 3b = the shifted ideal position SF
TDP3−actual position REALP3b <0, and new borrowing NEWDBT3b = −1 × index 3b. These functions are performed by the CDV calculation processing function cnt_clc_cdv_s
how.

Next, the allowable CDV value will be described. The initial value of the allowable CDV value is 7, and after that, E_CNT_CDV
E_CNT_CDV_AD for each break of _CHKi
The value of Di is added as shown in the table of FIG. This is an allowable CDV value setting function cnt_set_allow_
This is done by cdv. FIG. 45 is a flowchart showing the procedure of the allowable CDV value setting function cnt_set_allow_cdv. This function corresponds to the processing tr in FIG.
Used in unc_s3_s81.

However, the above-mentioned allowable CDV value is used as an allowable value of how far it can be shifted forward and backward when the idealized GID converted into an integer has already been arranged. The GID is not for use as an allowable value in a range deviating from the GID of the real number calculation. It is necessary to control the range in which the integerized GID deviates from the GID of the real number calculation to be less than one. This is because, when a measurement device is connected for observation, a position based on equal intervals in the calculation of the real number is used as a reference, so that there is a possibility that the CDV is detected as an error exceeding the reference value.

In the following, a description will be given of a random selection of an arrangement position using a random number in the present embodiment. First, a method of generating a random number sequence according to the present embodiment will be described.
In the random number generation algorithm according to the present embodiment, 15 types of random number sequences are prepared for each random number generation cycle. This is shown in FIG. The period of the random number generated by the random number sequence is a power of 2 number 2 ^p, and from 0 to (2 ^p −
A random number Ra in the range of 1) is generated. Generally from 0 to (G
When a random number Rg up to max-1) is obtained, conversion is performed by the following equation. Rg = Ra × Gmax / 2 ^p

In the random number generation algorithm according to the present embodiment, the maximum period of the random numbers that need to be generated is the total number of slots 11980 in the GID arrangement table, and 8192 = 2 ¹³ <11980 <2 ¹⁴ = 16384. The cycle is up to 2 ¹⁴ . However, a similar algorithm can generate a random number with a period exceeding this.

The function of cycle 1 shown in FIG. 46 always generates 0 and is not used as a random number, but is provided for consistency as a whole at the time of an error or the like. In the random number generation algorithm according to the present embodiment, the argument is p, that is, the logarithm cycle_l of the base 2 of the cycle cycle.
og2 is provided, and a function cnt_rand is provided which rewrites a random number sequence by calculating a predetermined variable of a random number table every time it is called. FIG. 47 is a flowchart showing cnt_rand.

This function corresponds to the process trunc_s in FIG.
1_s2 base offset setting, processing tru in FIG.
nc_s2_s1 start offset setting, processing trunc_s3_s3 random slot search in FIG. 12,
Processing cnt_chk_cdv_pat_val of FIG. 38
It is used in many places where random numbers are used, such as selection of a candidate position for ue_s10.

For initialization of the random number table, p as an argument, ie, cycle_log2, and an initial value se
There is provided a function cnt_slanc that gives ed and allows a predetermined variable to be initialized to seed. FIG. 48 is a flowchart showing cnt_sland. This function is used in the process trunc_s0_s2 of FIG.

Each time the function cnt_rand of FIG. 47 is called, it updates the random number value of the cycle cycle and returns a new value. The update formula of the process cnt_land_s1 and the initialization formula of the process cnt_land_s1 are based on the following formula (2). Meaning of _{x n + 1 = 1103515245x n +12345} (modulo2 (cycle_lo g2)) x0 = seed (2) This equation, starting from the initial value x ₀ = seed,
Sequentially 1103515245x _n +12345 2
^(Cycle_log2) ), and the remainder is _{xn + 1} .

Next, the basis of the above equation (2) will be described. For example, Kazumasa Wakimoto, "Knowledge of random numbers", Morikita Publishing
Elementary Information Processing Course 5,1981, pp. As described in No. 32, as a general method of generating a uniform uniform pseudorandom number by a computer, a mixed congruence method (mixed congrue method) is used.
nce method). This is a method of generating a random number sequence according to the following equation. x _{n + 1} = ax _n + c (modulo2 ^p ) x ₀ = seed The meaning of this expression is to start from an initial value x ₀ = seed,
That is, a remainder obtained by sequentially dividing ax _n + c by 2 ^p is ^defined as x _{n + 1} .

Here, the random number sequence x generated by the above equation
The maximum period of ₀ , x ₁ , x ₂ ,... Is a = 1 (modulo 4)... When it is divided by 4, the remainder is c1 (modulo 2). , ^2p . In the random number generation algorithm according to the present embodiment, B.V. W. Kerni
ghan and D.S. M. Ritchie, "The
C programming language S
second edition ", Prentice H
All, 1988, pp. Although the following values that are the same as the random numbers of 46 are used, other values that satisfy the above equation may be used. a = 1103515245 c = 12345

The random number generation algorithm itself in this embodiment is based on the mixed congruential method conventionally known in the above-mentioned literature and the like. However, in solving the CID placement problem, the random number generation period is very long. It is characterized by the fact that a random number function that can control the random number generation period according to the purpose is created by using a mixed congruence method instead of a built-in function of a programming language. A comparison between the random number algorithm of the present embodiment and a random number function of a built-in function of a programming language will be described below.

According to the random number algorithm of the present embodiment, if a random number having a period of 2 ^p is generated 2 ^p times, the random number is changed from 0 to (2
It has been confirmed that the number of ^p- 1) occurs exactly once each. Since the order of generation of a sequence within a cycle is determined by seed, if random numbers for two cycles are generated, the order of the series in the first cycle and the second cycle becomes equal. Function rand for finding random numbers in the standard library of the programming language C
Exists, but the C rand function is created so that the cycle of random number generation is long, and the fineness of random number generation cannot be specified. Therefore, a random number of a number between 0 and (2 ^p -1) is set to 2 Even if it occurs ^p times, there are many such numbers that never occur or that occur multiple times. Since this does not meet the purpose of the present embodiment, a random number function is handmade.

FIG. 49 shows a comparison between the random numbers according to the present embodiment and the built-in random numbers of Visual C ++ by Microsoft Corporation, which is a C language software that has been simulated. The solid line in FIG. 49 was obtained as follows. First, a random number Ra between 0 and 16383 is generated as seed = 1 and a period 2 ¹⁴ as a random number according to the present embodiment, and this is converted into a random number R1 between 0 and 11979 by the following equation (3). R1 = Ra × 11980/16364 (3)

When this trial is repeated 11980 times,
Numbers between 0 and 11979 may occur once, occur twice, or never occur. Each number is indicated by a solid line in FIG. The number of occurrences exactly once is 7,188, and the number of occurrences exactly twice or zero is equal to 2,396. Although not shown in the figure, the same trial is repeated 16384 times.
Since Ra is generated exactly once every number between 0 and 16383, the number between 0 and 11979 of R1 is 7580 once, 4400 twice, and R1 Becomes zero.

The broken line in FIG. 49 is Microsoft Vis.
This is the frequency of occurrence of a number between 0 and 11979 when the trial for obtaining a random number between 0 and 11979 is repeated 11980 times using the built-in random number rand of ual C ++. The conversion formula is as follows. R3 = Rc × 11980 / (RAND_MAX + 1) However, in Microsoft Visual C ++, which is C language software that has been simulated, R
AND_MAX = 32767 = 2 ¹⁵ −1. There are 4449 occurrences that never occur, 4361 occurrences that occur exactly once, 2168 occurrences that occur twice, 773 occurrences that occur three times, 182 occurrences that occur four times, and five occurrences The number of occurrences is 46, and the number of occurrences 6 is 1.
It can be seen that the built-in random number C has a non-uniform frequency of random number generation as compared with the random number according to the present embodiment, and is not suitable for generating a number within a certain interval.

In the solid line data of FIG. 49, the number of occurrences is 71.
There are 88. FIG. 50 shows how this value changes when the seed of the random number is changed.
In the present algorithm, the seed is changed between 0 and 11979, and in each case, a random number with a period of 16384 is generated 11980 times, and in the above equation (3), 0 to 1197
When converted into a random number between 9, the number of numbers that occur exactly once is plotted on the horizontal axis, and the frequency of occurrence is plotted on the vertical axis. Regardless of the number of seeds, it can be seen that the number of occurrences falls between 7120 and 7320.

Functions related to random numbers include the following. FIG. 51 shows a function cnt_l for calculating the logarithm of the base 2
6 is a flowchart showing og2. This function is shown in FIG.
1 process cnt_chk_cdv_rand_s1. FIG. 5 shows a procedure for calculating the logarithm of base 2.
1 will be described. When the function cnt_log2 is called together with a variable y as an argument, x that satisfies 2 ^x ≦ y <2 ^{(x + 1)} is obtained for the argument y in cnt_log2_s1, and the obtained x is returned with cnt_log2_ed. When a value within the range of the added y is calculated by using a random number, the logarithm of 2 (log2) of the cycle is calculated by rounding down. CD
V allowable random number GID search processing function cnt_chk_
Used in cdv_rand.

In addition, the random number G within the CDV allowable range shown in FIG.
ID search processing cnt_chk_cdv_rand, FIG.
GID random number search processing of 0 (for all GIDs) function cnt_
chk_start_rand_all, and cn
t_chk_start_rand_all_delt
a uses a random number. These have been described above.

By the way, in the above description, the present invention relates to A
The case of using for the GID placement problem in the GFR of the TM has been described. However, G shown in FIG. 1 or FIG.
Similarly to the ID table GIDTBL, the slot SLOT having the address ADR has a group having the attribute shown in FIG. 3, that is, a group having the number of elements of GPCR, a group identifier of GID, and a tolerance of a deviation from an ideal position of one or more. It goes without saying that the kind of group element can be used for other problems with the purpose of evenly distributing.

As described above, according to the first embodiment, when the arrangement processing time is short and the arrangement cannot be performed at the ideal arrangement position, an appropriate alternative position is selected, and the distributed arrangement is uniformly performed as a whole. Is obtained.

Embodiment 2 In the second to fourth embodiments, the function cnt_mix_rea shown in FIG.
1). Function cn of Embodiment 1
t_mix_real is the total number wh to be divided
In the slot of ole_int, the division number divider_
It is a function that performs the process of arranging the int elements as evenly as possible in the real world, writes the integer value to the arrangement table, and returns it. The input and output from the outside is an integer, and the real number calculation is performed only in this function. It was meant to be closed.

Function cnt_ shown in FIG. 15 of the first embodiment
In mix_real, processing cnt_mix_real
_S1 and integer variables whole_int and divider
_Int is a single precision floating point variable whole and div
Convert to ider and process cnt_mix_real_s
In step 4, the value of the above equation (1) was calculated. In this embodiment, the following equation (4) is used as the distribution position calculation equation corresponding to equation (1). answer = whple * i / divider (4) However, the value of each parameter is as shown in FIG.

FIGS. 53 and 54 show the method of calculating the answer in this embodiment. Processing cnt of these figures
_Mix_real_s1 to cnt_mix_rea
l_s4 corresponds to the same process in FIG. FIG. 53 and FIG.
4, other processes in FIG. 15 are omitted. FIG.
In the answer calculation procedure shown in FIG.
mix_real_s1 is an integer variable whole_in
t and divider_int are converted to double-precision floating-point variables w
It is converted into hole and divider.

Using the equation (4), i = 1, 2,..., Div
One of the methods of sequentially calculating the answer for each ider is a method of first multiplying whol and i by the following formula for each i, and then dividing by the divider. answer = (whole * i) / divider (5)

This corresponds to the processing cnt_mix_re shown in FIG.
al_s4. Here, if whole and i take the values shown in FIG. 52, the product becomes the maximum. By the way, in a normal floating point display conforming to the IEEE standard, the number of significant digits of the decimal display is 6 digits in float, and
le has 15 digits. Therefore, the calculation of equation (5) is f
In the case of "loat", the number of significant digits exceeds the number of significant digits, and a digit loss occurs, so that a valid value cannot be obtained. On the other hand, when double is used as shown in FIG. 53, an effective value can be obtained without occurrence of digit loss.

Another method of calculating the expression (4) is a method of first dividing the whole by a divider and then sequentially multiplying each i. answer = (whole / divider) * i (6) With this method, cnt_mix_real_ in FIG.
If the division is performed once in s4_1 first, in each loop in which i is changed, only the multiplication as in cnt_mix_real_s4_2 may be performed, and the calculation speed is increased. In addition, since division is performed first to make a small number, overflow does not occur. However, in the case of the float, a valid value cannot be obtained because of the insufficient number of significant digits in this case as well.

As an example, divider = 380, i =
Taking the case of 19, whol / divider = 111980/380 = 3
1.5263 answer = (whole / divider) * i = 31.5263 * 19 = 598.9997 ≒ 599.
000, where an an as a float variable
The content of the switch is in decimal notation, +5.989999998
092651367e + 0002. This is converted to an integer of 598, and the desired 599 cannot be obtained. Also in this case, as shown in FIG.
If e is used, a valid value can be obtained.

As described above, according to the second embodiment, the double-precision variable doubl
By using e, it is possible to obtain an effect that an effective value can be obtained without occurrence of a digit loss.

Embodiment 3 Embodiment 3 relates to improvement of the function cnt_mix_real shown in FIG. 15 of Embodiment 1. In the function cnt_mix_real shown in FIG. 15 of the first embodiment, the processing cnt_mix
_Real_s1 and integer variables whole_int and di
vider_int is converted to floating point variables whole and di
is converted into a “vider” and the process is cnt_mix_real_
At s4, the value of equation (1) was calculated. In this embodiment, (4) of the second embodiment is used as the distribution position calculation formula corresponding to (1) of the first embodiment, and the values of the parameters are as shown in FIG. In this embodiment, (4)
Use fixed-point variables in formula calculations.

Prior to showing the procedure for calculating the arrangement position according to the present embodiment, a procedure for performing calculation using a general fixed-point display will be described with reference to FIG. Here, since the original number of the normal display is an integer, the decimal point is on the right side of the least significant bit. However, if the original number is a decimal, it is only necessary to place the decimal point at the corresponding position. The procedure is to first convert the integer in the normal display to the fixed-point representation by shifting the desired number of bits,
The necessary calculations are then performed, and the result is shifted back to the normal display number. Number of bits to shift SHIFT
Depends on the required accuracy. Variables in fixed-point notation must have a bit length that does not overflow when shifted.

FIG. 56 shows the first a in this embodiment.
The calculation method of nswer is shown. Processing cnt_ in these figures
mix_real_s1 to cnt_mix_real
_S4 corresponds to the same process in FIG. In FIG. 56, other processes in FIG. 15 are omitted.

A method of multiplying the expression (4) by first multiplying the whole by i and then dividing by a divider is shown in the following expression (7) and FIG. whole = whole << SHIFT divider = divider << SHIFT answer = (whole * i) / divider (7) where whole << SHIFT indicates that the integer whole is shifted SHIFT bits to the left.

This corresponds to the processing cnt_mix_re in FIG.
It is indicated by al_s1. First processing cnt_mix
_Real_s1 left shift and in
t and the fixed-point representation are obtained, and i = 1, 2,.
The answer is calculated sequentially for each of the providers. In this method, since the left shift is canceled by the numerator and the denominator of the equation (7), the return shift by the right shift is not necessary.

FIG. 57 shows the required number of bits at this time.
It is. Since whole and i take values in the range of FIG. 52, the required number of bits is 14 bits and 13 bits, respectively. In addition, “divid” takes a value in the range of FIG.
In order to obtain a valid result by performing division by er, 13 shift bits are required. These add up to 40 bits.

Microsoft used in the simulation
In ft Visual C ++, the maximum number of bits is an unsigned 32-bit integer.
ng, and the maximum value is 4294967295 = 2 ³² −1), so that the simulation using the 40-bit display shown in FIG. 57 cannot be performed entirely. 11980 × 2 ¹³ × 44 = 4318167040> 4294967295 = 2 ³² −1 11980 × 2 ¹³ × 43 = 4220026880> 4294967295 = 2 ³² −1 (8)

However, as shown in the above equation, when i does not exceed 43, 32 bits are sufficient for the fixed-point display variable. Then, as a result of an experiment in which the divider was in the range of 2 ≦ divider ≦ 43, a valid result without errors could be obtained. Therefore, if an integer of 40 bits or more can be prepared as a fixed-point display variable, an effective result can be obtained by this method.

FIG. 58 shows the second a in this embodiment.
The calculation method of nswer is shown. This is described for comparison with the first method described above, and is inferior in performance to the first calculation method.

First, the calculation of equation (4) is performed by
A method of performing division by a vider and then sequentially multiplying each i is shown in the following equation (9) and FIG. whole = whole << SHIFT whole_divide = whole / divider answer = whole_divide * i answer = answer >> SHIFT (9)

The left shift of the first integer whole is shown in FIG.
Calculation by cnt_mix_real_s1
The division by ole / divider is cnt_mix_
In real_s4_1, it is calculated in advance and the integer who
stored in le_divide, i = 1, 2,.
cnt_mix_real_s4 per divider
_2 to sequentially calculate the answer, and finally cnt_mi
In x_real_s4_3, the number is returned to the normal display number by a right shift.

In this method, if division is performed once before, in each loop where i is changed, only multiplication may be performed. However, in this method, cnt_mix_real_s4
A truncation error occurs when calculating the integer whole_divide of _1, and the subsequent multiplication by i cannot recover the truncated portion, so that a valid result cannot be obtained. In some cases, you can get the same exact value as the real number calculation, but the whole and the divider
When the relationship between i and i is exactly divisible, a valid value cannot be obtained.

As an example, divider = 380, i = 1
Consider the case of 9. Substituting into the equation (4), anser = whole × i / divider = 111980 × 19/380 = 599 (10), which is exactly divisible.

On the other hand, substituting SHIFT = 13 into equation (9) gives the following. whole = whole << SHIFT = 11980 << 13 whole_divide = whole / divider = whole / 380 answer = whole_divide * i = whole_divide * 19 answer = 13

[0175]

(Equation 1)

The result of equation (11) is different from that of equation (10). Even if the shift bit SHIFT is further increased,
Similar results are obtained. The fundamental problem with this method is that integer division always truncates the quotient if it cannot be divided. For example, if the division and the multiplication are performed at the same time as in the equation (10), the division cannot be recovered by the subsequent multiplication as long as the division is performed first and the quotient is obtained by the truncation in the calculation that is exactly divisible. Therefore, even if the number of shift bits is increased, a valid result may not be obtained by this method. By the way, in the calculation of the floating-point representation, such a problem does not occur because the number is rounded to the nearest when rounded.

FIG. 59 shows the third a in this embodiment.
The calculation method of nswer is shown. This is described for comparison with the first calculation method, and the performance is inferior to the first calculation method.

The problem with the second calculation method is that the first whol
In the division of e by the divider, the quotient was rounded down to an integer. Then, if only the first quotient is calculated as a real number and the result is used for subsequent multiplication,
I think it will work. As described in the second embodiment, since it was found that a float is not enough for a real number, double is used. The procedure is shown in the following equation (12) and FIG. Do to clarify the attributes of the variable
"uble" has db and "integer" has integers.

[0179] whole_divide_db = whole_db / divider_db; &#9; / * The point is that not truncated in this division * / whole_divide_db_shift = whole_divide_db; for (p = 1; p <= SHIFT; p ++) {whole_divide_db_shift &#9; = whole_divide_db_shift * 2;} / * Real SHIFT bit shift * / whol_divide_db_shift_int = (unsigned long) whole_divide_db_shift if; Processing similar to equation (9) using integers * / answer_int whole_divide_db_shift_in t * i; answer_int = answer_int >> SHIFT; (12)

In this method, as described above, a 32-bit unsignedlong is used as an integer, but unfortunately, the validity cannot be verified with a 32-bit integer. Using this method requires an integer length of at least 64 bits.

As described above, the first calculation method is effective in this embodiment, and the arrangement position an by the fixed-point notation is displayed.
In order to obtain the switch, as shown in FIG. 56, it is necessary to perform division after multiplication to make the integer 40 bits or more.

As described above, according to the third embodiment, by using fixed-point variables in the real number interval calculation processing, it is possible to perform processing using only integers and to shorten the arrangement processing time. The effect is obtained.

Embodiment 4 Fourth Embodiment
This is related to the method of the real number interval calculation processing of, that is, the improvement of the function cnt_mix_real shown in FIG. 15 of the first embodiment. Function cn shown in FIG. 15 of the first embodiment
In t_mix_real, processing cnt_mix_re
In al_s1, integer variables whole_int and div
der_int is replaced with floating point variables whole and div
Convert to ider and process cnt_mix_real_s
In step 4, the value of equation (1) was calculated. In the third embodiment, (4) of the second embodiment is performed using fixed-point display variables.
The value of the equation was calculated, and the value of each parameter was in accordance with FIG.

In the third embodiment, the calculation method using the fixed-point display variables has been described. In the calculation using the fixed-point display variable, as described in the first method of the third embodiment, division is performed after multiplication, and it is shown that an integer requires 40 bits or more. However, since integers of 32 bits or more are not standardized and depend on the compiler, it is desirable to avoid using them if possible to prevent malfunction. Therefore,
In this embodiment, a method of performing fixed-point calculations with integers of 32 bits or less will be described.

In this embodiment, the formula 1 of the first embodiment is used.
Is used as the distribution position calculation formula corresponding to (2), the values of the parameters are as shown in FIG. 52, but the number of bits to be shifted is represented by S. In this embodiment, a fixed-point calculation is performed by using an integer of 32 bits or less in the calculation of Expression (4).

Prior to showing the arrangement position calculation procedure according to this embodiment, the derivation of the calculation method in this embodiment will be described. The calculation method shown in this embodiment is represented by res_re
Called the s method. This res_res is a “residue”
of residue, that is, the remainder.

As a first method of this embodiment, first, consider the case where the rounding bias shown in the first embodiment is set to zero. The calculation policy is as follows.・ To shift the bit, the whole is divided
Who is the residue when divided by
w that is a quotient with only le_res
hole_quo does not shift.・ Transform the expression by equivalent conversion as much as possible. (If there is the same number in the denominator and the numerator, delete them, etc.) ・ Perform multiplication first. • Division with truncation • Bit shift is not performed to the end.

The constant “whole” and the variable “divide”
The following is a modification of the equation for obtaining an answer from r.

[0189]

(Equation 2)

The first term on the right side of the above equation (whole / div
The quotient of the ider is calculated by multiplication without performing the bit shift, and the second term (the remainder) is calculated by performing the bit shift.

In the following, whole = 11980, div
Numerical example when ider = 380, i = 19, S = 13
Is shown. 11980 = 31 × 380 + 200 11980 × 2¹³= 31 × 2¹³× 380 + 200 × 2¹³  Where 200 × 2¹³= 200 × 8192 = 4311 × 380 + 220 Therefore, 11980 × 2¹³= (31 × 2¹³× 38) + (4311 × 380 + 220) = (31 × 2¹³+4311) × 380 + 220 answer × 2¹³= (11980 × 2¹³× 19) / 380 = {(31 × 2¹³+4311) × 380 + 220} × 19/380 = 31 × 2¹³× 19 + 4311 × 19 + (220 × 19) / 380 = 31 × 19 × 2¹³+ 4311 × 19 + 11 = 31 × 19 × 2¹³+ 81909 + 11 = 31 × 19 × 2¹³+ 81920 = 31 × 19 × 2¹³+ 10 × 2¹³  = 589 × 2¹³+ 10 × 2¹³  = 599 × 2¹³  answer = 599

A program example of the above calculation is shown below.
A flowchart is shown in FIG. whol_quo = whole / divider; whol_res = whole% divider; res_s = whole_res <<S; res_quo_s = res_s / divider;
i; res_res_ans_s = res_res_s * i
/ Divider; res_ans_s = res_quo_ans_s + r
es_res_ans_s; res_ans = res_ans_s >>S; quo_ans = whole_quo * i; answer = quo_ans + res_ans;

Hereinafter, the required number of bits in the first method of this embodiment will be examined. Fig. 61 explains each variable.
Shown in The range of values for each variable is shown in FIG. 61, but these values vary depending on the value of the divider. Therefore, simply adding the required number of bits in the value range of FIG. 61 results in a value exceeding the really required number of bits. In order to correctly evaluate the required number of bits, it is necessary to investigate the relationship between the maximum value of each variable and the value of the given divider.

FIG. 62 shows the relationship between the given divider value and the maximum or upper limit value of each variable.
In FIG. 62, the shift number S is 13. Di on the horizontal axis
The maximum value of the value of the “vider” is actually 5990, but is shown up to 11980 so that the behavior of each variable can be clearly understood. The vertical axis is the log2 of each variable rounded up and indicates the number of bits required to represent each variable.
From FIG. 62, it can be seen that when divider ≦ 5990 and S = 13, the required number of bits is 26 bits.
By generalizing this, it seems that the number of required bits may be set to S + 13. Therefore, if S ≦ 19, a 32-bit integer can be used.

Next, a case where a rounding bias is added will be described as a second system of the present embodiment by extending the first system of the present embodiment. Rounding bias (ro
under bias) is a run that takes a value from 0 to 99
d_bias_int and round_bias_pit
ch = 100, round_bias = (round_bias_in
t) / (round_bias_pitch), and takes a value from 0 to 0.99.

In the first embodiment of the present embodiment, the expression ans
wer is as follows.

[0197]

(Equation 3)

An example of the program is as follows, and its flowchart is shown in FIG. whol_quo = whole / divider; whol_res = whole% divider; res_s = whole_res <<S; res_quo_s = res_s / divider; res_res_divers_res = diversizer
int << S) / round_bias_pitch; / * Without parentheses in the above expression, division is performed first.
* / &#9;&#9; / * Calculation not dependent on i so far.//res_quo_ans_s=res_quo_s*
i; res_res_ans_s = res_res_s * i
/ Divider; res_ans_s = res_quo_ans_s + r
es_res_ans_s + round_bias_
s; res_ans = res_ans_s >>S; quo_ans = whole_quo * i; answer = quo_ans + res_ans;

The number of bits required for adding a rounding bias will be described below. In the first method of this embodiment, if the rounding bias is not added, the required number of bits is S
+13. Run when shift bit S = 13
FIG. 64 shows a change in the value of d_bias_s with respect to round_bias_int. From FIG. 64,
It can be seen that round_bias_s requires 13 bits. This depends on S. When the rounding bias is added, as described above, res_ans_s becomes ro
und_bias_s.

FIG. 65 is obtained by adding the rounding bias of FIG. 64B to FIG. 62E. This is shown in FIG.
(E), and as in FIG.
When der ≦ 5990 and S = 13, it is considered that the required number of bits may be S + 13. Therefore, if S ≦ 19, a 32-bit integer can be used.

The evaluation of the first method of this embodiment will be described below. The evaluation was performed by changing round_bias from 0 to 0.99 in increments of 0.01 in the case of shift bits S = 0, 9, and 13, and examining the number of evaluation indexes for each value. For each value of round_bias, the band (divider) was changed from 1 to 5990, and for each band, i was changed within the range of 1 ≦ i ≦ divider.

FIG. 66 shows the evaluation indices and the results. step
Since the interval between 1 and step 2 does not differ from the specified value obtained by the integer calculation in any of the three cases, it does not seem that a troublesome problem such as a change in the band will be caused no matter how many shift bits are used. It is. However, in comparison with the answer by double which gives an ideal value, the smaller the shift bit, the smaller the answer
When the numbers are different, the number increases. an by double
If a value different from "swer" is taken, the CDV increases to a maximum of 1, so if the CDV check is performed immediately after that, an error may occur.

By the way, from (b) and (c) of FIG. 67,
0, 0.25, 0.5, 0.75 when S = 9 and 13
In the four values, the difference from double is zero. Therefore, if these four values are used as round_bias, it can be said that there is no difference from double. FIG.
7 (a), there is a point where the difference is small even when S = 0, but even then, there are about 1000 differences except for the point where round_bias = 0. round_b
When ias = 0, all the evaluation indices in FIG. 66 became zero when S = 0, 12, 20, 21.

In this embodiment, a method of realizing the calculation of the ideal arrangement position of the GID arrangement problem by using a fixed point with an integer of 32 bits or less was examined, and the res_res method was derived. Regardless of whether the rounding bias is added or not, when the shift bit is S, the required number of bits may be S + 13. Therefore, if S ≦ 19, a 32-bit integer can be used.

In the evaluation by the simulation, in the case where the shift bits S = 0, 9, and 13, round_round
When bias was changed from 0 to 0.99 in increments of 0.01, the numbers of step1 and step2, the total number, and the number of intervals other than step1 and step2 were zero. In comparison with the calculation using double, which is an ideal value, the smaller the shift bit, the larger the number of cases where the answer is different. As a result, the CDV increases to a maximum of 1, and an error may occur if the CDV check is performed immediately thereafter. However, since the specified value (the number of steps, etc.) obtained by the integer calculation is maintained, it is unlikely that a troublesome problem such as a change in the band will be caused.

As described above, according to the fourth embodiment, in the real number interval calculation processing, the calculation using the fixed-point variable can be performed with an integer of 32 bits or less, and the arrangement processing time can be shortened. Is obtained.

Embodiment 5 FIG. 68 is a schematic diagram showing a configuration of an apparatus for realizing a shaping function according to the fifth embodiment of the present invention. In FIG. 68, reference numeral 1 denotes a CPU for controlling a shaping circuit or the like, and 2 controls output traffic by receiving an input of a cell stream of ATM cells (fixed length cells), adjusting the interval between cells, and outputting the adjusted data. The shaping circuit 3 reads the header information (address information such as VPI and VCI) of the transferred ATM cell, and
A switch unit, which is provided as a routing switch for sending an ATM cell to an output interface unit of an output line to which a TM cell is to be transferred, 4 is an input interface unit for realizing an interface function between the input line and the switch unit 3, and 5 is An output interface unit that realizes an interface function between the output line and the switch unit 3.

The switch unit 3 includes an ATM cell aggregating unit for concentrating transmission lines extending from the plurality of input interface units 4, collecting ATM cells transferred from each input interface unit to form an ATM cell stream, and an ATM cell aggregation unit. A sequentially from the ATM cell stream configured by the cell aggregation unit
An ATM cell distributing unit is provided for extracting a TM cell and distributing each ATM cell to a desired output interface unit based on header information.

[0209] The shaping circuit 2 is provided with the A
It is arranged between the TM cell aggregator and the ATM cell distributor. Thus, assuming that the transfer rate of the ATM cell on the input line or the output line is about 150 Mbps, the shaping circuit is provided with an output table having a bandwidth many times that of the shaping circuit in accordance with the degree of concentration in the switch section.

The maximum output cell band (Peak Cell) related to the transfer rate of the ATM cell in each connection.
When setting l Rate (hereinafter referred to as PCR), a PCR may be calculated for one connection (VC connection, VP connection, etc.), or a connection group configured as a set of several connections. It is also conceivable to calculate PCR in units of. In the following description, the present invention will be described on the assumption that the PCR is set for each connection group.

Regarding connection grouping, A
A plurality of connections having the same bandwidth are grouped together based on a communication report from a user to a connection admission control unit of a network when setting up communication using TM, and a predetermined bandwidth is used. And collectively grouping a plurality of connections that can be appropriately grouped as one group, and a plurality of connections that can be grouped based on predetermined processing by the CPU. Etc. are considered.

FIG. 69 shows an output table (group identifier arrangement table or GI) set in the shaping circuit.
FIG. 14 is a diagram illustrating an example of a (D table). 10 is the bandwidth 1.2
Gbps, output table 11, output table 1
A unit table (slot) 12 obtained by dividing 0 into units of about 0.1 Mbps is a number on the output table 10, that is, an address (address) in arranging a connection group. In this example, 1.2 Gb
The ps output table 10 is divided into about 0.1 Mbps unit tables 11, and 11980 unit tables 11
Is generated. And 1.2 Gbps is given as an indication of the reading speed of the ATM cell by the routing switch.

[0213] The unnumbered unit table 11 in the output table 10 indicates that the unit table 11 is empty, and if the unit table 11 is numbered, the connection group of that number indicates that the connection group has that number. This indicates that it is arranged in the unit table 11. Unit table 11
If 100 is added to the unit table, the unit table 1
1 indicates that the connection group 100 is allocated, and n indicates the ideal transmission interval of the connection group 100.

The individual unit table 11 contains a large number of ATMs transferred through the grouped connections.
Identification information about the cell is properly accommodated. The transmission of the ATM cell is sequentially performed for each unit table 11, and
From the reading of one unit table 11, one AT
M cells are transmitted. When the last address of the output table 10 is reached, the address returns to the initial address (that is, the output table number “0”), and similarly, the next ATM cell arranged for each unit table 11 is transmitted.

Next, the operation for performing shaping will be described. FIG. 70 is a flowchart showing a shaping method according to the fifth embodiment of the present invention. The value of the PCR of each connection group is determined by, for example, the shaping circuit 2
May be determined by the CPU 1 controlling the CPU 1 or may be set in advance and stored in a memory associated with the CPU 1.

First, in process ST20, PCR (band value)
Sort the connection groups in descending order of the connection groups. When sorting by sorting is performed, the process S
At T21, the number of connection groups having the same PCR is counted for each PCR. Next, in step ST22, an ideal cell transmission interval n of the processing target group is calculated. Processing S
In T23, it is determined whether or not a plurality of connection groups including the same PCR as the processing target group exist, including the processing target group. If there is no other group having the same PCR, the processing branches. And process S
Move to T24.

In the process ST24, the first start address at which the processing target group is arranged on the output table is calculated by using a calculation method such as a random number. With respect to the head arrangement address (arrangement start position) calculated in this way, it is determined whether or not the processing target group can be arranged in processing ST25. If another connection group has already been allocated, the process returns to step ST24 and the calculation is performed again. Conversely, if the arrangement is possible, the process ST2
In step 6, the address is arranged at the corresponding start location on the output table. Next, in the process ST27, the processing target groups are sequentially arranged within the allowable CDV value based on the head arrangement address. More specifically, an operation of (head arrangement address) + N * n is performed to determine an ideal arrangement address (ideal arrangement position), and whether or not it is possible to arrange at the ideal arrangement address is determined. , A neighborhood is searched within a range within the allowable CDV value. In the above equation, N indicates a natural number from 1 to the total number of arrangements for the processing target group.

Regarding the above neighborhood search, 1) search within the range of allowable CDV value; 2) minimization of deviation from ideal arrangement address; 3) unit in which another connection group has already been arranged It is possible to adopt an algorithm that determines based on three conditions expressed as follows: For example, three conditions are expressed as parameters, and each parameter is weighted.
An algorithm that considers an address that minimizes the sum of the parameter values as an arrangement address is considered.

[0219] Further, regarding the search of the arrangement address, not only the ATM cell is shifted backward as in the prior art, but also
An empty address search is performed both in the forward and backward directions.
Here, in the arrangement of the unit tables, it is required to avoid being adjacent to each other because, when adjacent, the collision is likely to occur at the time of arranging the next connection group (arrangement of ATM cells), and the output is also reduced. This is because congestion easily occurs in the interface unit. Then, in the process ST28, it is determined whether or not the arrangement of all the connection groups has been completed. If there is any remaining, the process returns to the process ST22 to continue the process. If there is no remaining, the process is completed and the setting of the output table is completed. I do.

If the number of connection groups having the same PCR in the process ST23 is two or more, the process branches to the process ST29, and the process target group is arranged on the output table as in the process ST24. The head address is calculated using a calculation method such as a random number. With respect to the head arrangement address calculated in this way, the process ST3
It is determined whether or not it can be arranged with 0, and if it is already arranged, the process returns to the process ST29 to perform the calculation again. If it can be arranged, the process target group is arranged at the head arrangement address in the process ST31. I do.

Next, in processing ST32, assuming that the number of connection groups having the same PCR is m, between the head arrangement address determined in processing ST31 and the ideal transmission interval n, the head arrangement address of each group is determined. Set evenly. In the process ST33, it is determined whether or not each connection group can be arranged at the head arrangement address.

If there is a head arrangement address that has already been arranged (cannot be arranged), for the head arrangement addresses of those connection groups, a head arrangement address determination process by neighborhood search is performed in process ST34. In this neighborhood search, the condition that the search is performed within the range of the allowable CDV value is not imposed as in the process ST27, but, for example, 1) the deviation from the uniformly arranged address is minimized, 2) It is possible to adopt an algorithm that is determined based on two conditions expressed as follows, in which another connection group is not adjacent to the unit table in which the connection table is already arranged as much as possible.

Next, in process ST35, for each connection group, the number of successively multiplying the ideal transmission interval n by a natural number (N) from 1 to the total number of arrangements is obtained by adding the top arrangement address of the processing target group. The ideal arrangement address is calculated, and if the arrangement is possible at the ideal arrangement address, the arrangement is performed. If the arrangement is impossible, the neighborhood search similar to that performed in the process ST27 is performed to determine the arrangement address.

When the processing is completed for one processing target group, it is determined in processing ST36 whether or not the arrangement has been completed for all the connection groups having the same PCR. ST35
The processing is returned to the next connection group having the same PCR, and if completed, the processing S
Proceed to T28. In the process ST28, it is determined whether or not the process has been completed for all the connection groups. If there is any remaining, the process returns to the process ST22 to continue the process. Complete the process.

As described above, according to the fifth embodiment, for a plurality of connection groups having the same PCR, the head arrangement addresses of the respective connection groups are equally set, and the plurality of connection groups having the same PCR are set. Are arranged so as to be evenly distributed from the top address to the last address, so that the connection groups can be more evenly arranged on the output table, preventing collision and congestion of ATM cells. Therefore, an effect is obtained that a larger number of connections can be secured for the allowable CDV value.

If the connection to be processed cannot be located at the ideal location address, an empty address ahead and behind is detected within the allowable CDV value centered on the ideal location address to determine the location address. With this configuration, empty addresses in the range of the allowable CDV value can be used efficiently, so that a larger number of connections can be secured for the allowable CDV value and collision of ATM cells frequently occurs. Even in such a case, an effect is obtained that the output band is not reduced unlike the search for only the backward.

Further, when there are a plurality of addresses in which the connection to be processed can be arranged within the range of the allowable CDV value, the adjacent front and rear addresses are not already arranged by other connections. Can be preferentially selected and the connection to be processed can be arranged, so that the arrangement of connections at adjacent addresses can be prevented as much as possible, and the collision of ATM cells can be reduced. Thus, an effect that a larger number of connections can be secured can be obtained.

Further, since a wide-band output table having a capacity sufficient to define the output order of ATM cells transmitted to all output interface units is provided, the arrangement of connection groups on the output table is slightly modified. And the like can be easily implemented, so that the effect of facilitating the creation of the output table and improving the efficiency can be obtained. Further, by preparing such a wide-band output table, the adjacency of the address where each connection group is arranged is prevented as much as possible.
The effect is obtained that a larger number of connections can be secured for the value.

In the fifth embodiment, a wide-band output table is provided for the switch unit. However, such a wide-band output table may be provided for each output interface unit. It is.

Embodiment 6 FIG. An apparatus configuration including a shaping circuit, a routing switch, and the like for realizing the shaping method according to the sixth embodiment, and a wideband output table used for shaping are shown in FIG.
69 are the same as those shown in FIG. 69, and therefore, the same symbols are used and their detailed description is omitted here.

Next, the operation will be described. FIG. 71 is a flowchart showing a shaping method according to the sixth embodiment of the present invention. First, in process ST40, the connection groups are sorted in the descending order of the PCRs (descending order), and the connection groups are rearranged in the descending order of the PCRs. Next, in process ST41, zero is set as an initial value of the total bandwidth value PCR_SUM (the total allocated bandwidth value) of the PCRs of the allocated connection group.

Then, the processing target groups are determined in the descending order of the bandwidth value, and for the specified processing target group, first, in processing ST42, the ideal cell transmission interval n of the processing target group is calculated and the number of failures is calculated. Set zero as the initial value. In the process ST43, the starting arrangement address for arranging the processing target group on the output table is calculated by using a calculation method such as a random number. In addition,
The head arrangement address is given as an initial value of the ideal arrangement address in the process ST44 for determining whether or not the arrangement is possible at the ideal arrangement address.

In the process ST44, it is determined whether or not it is possible to arrange at the calculated ideal arrangement address.
At T45, it is arranged at the corresponding address on the output table.
If the arrangement is impossible, the process proceeds to step ST46 to execute arrangement address search processing 1.

FIG. 72 is a flow chart showing the contents of the arrangement address search processing 1. In the arrangement address search process 1, first, in process ST101, it is determined whether or not PCR_SUM is equal to or smaller than a half value (judgment value) of the entire output band value (1.2 Gbps in the sixth embodiment). If it is less than half, the process proceeds to process ST102 to allow the allowable CDV.
Searches for an empty address among the odd addresses in the value. If the PCR_SUM is larger than half of the total output band value in the process ST101, the process proceeds to the process ST103 to allow the allowable CDV.
Searches for an empty address among even addresses in the value.

When there are a plurality of such vacant addresses irrespective of an even address and an odd address, as in the case of the neighborhood search in the fifth embodiment, for example, 1) a search in a range within the allowable CDV value is performed. Based on the following three conditions: 2) minimizing the deviation from the ideal allocation address; and 3) minimizing the proximity of the unit table in which other connection groups have already been allocated. It is conceivable to determine an optimal arrangement address.

After processing ST102 and processing ST103, the processing proceeds to processing ST104, and processing ST104 is performed.
Then, it is determined whether or not it can be arranged on the output table. If it can be arranged on the output table, the process ST107
Then, the process is terminated as “placeable”. If it cannot be arranged on the output table, the process S
Proceeding to T105, all empty addresses in the range within the allowable CDV value are searched. That is, in the process ST102, the allowable CD
If an odd address in the V value is searched for an empty address and placement is not possible, an empty address is searched for in an even address within the allowable CDV value, and in step ST103, an empty address is searched for in an even address within the allowable CDV value. If it is not possible to arrange, an empty address is searched for at an odd address within the allowable CDV value.

Then, based on the search result of process ST105, it is determined in process ST106 whether or not the output table can be arranged. If it can be arranged on the output table, the process proceeds to step ST107, where "arrangement is possible", and this processing ends. If it is not possible to arrange on the output table, the process proceeds to step ST108, and the process is terminated as "arrangement impossible".

Returning to FIG. 71, when the process ST46 is completed, the process proceeds to a process ST47, where it is determined whether or not the result of the layout address search processing 1 can be arranged. If the arrangement is impossible, the process proceeds to processing ST48, and it is determined whether or not the arrangement address is the head arrangement address. If it is determined in the processing ST48 that the address is the head arrangement address, the processing ST48 is executed.
49, the number of failures is incremented by 1, and then the process ST4
Then, the process returns to 3 and the calculation of the head address is performed again.

If it is determined in the process ST48 that the address is not the head arrangement address, the process proceeds to a process ST50 to determine whether or not the number of failures is less than the allowable search failure frequency. If the number is equal to or more than the allowable number of failures, abnormal termination (process ST51)
And the arrangement processing on the output table ends. If the number of failures is less than the allowable number of failures, the process proceeds to step ST52, where the number of failures is incremented by one, and the already allocated address of the processing target group is cleared and returned to an empty address. The process is performed again from the calculation of the address.

If it can be arranged in the process ST47, the process proceeds to the process ST45 to arrange the processing target group at the corresponding address on the output table. When the processing in step ST45 is completed, it is determined in step ST53 whether the placement of the processing target groups has all been completed. If not completed, the next ideal layout address is calculated from the formula {(starting location address) + n * N (N: a natural number from 1 to the total number of locations)} in processing ST54, and the process returns to processing ST44. Thereafter, the same processing is repeated.

If it is determined in the processing ST53 that all the arrangements of the processing target group have been completed, the processing ST53 is executed.
At 55, the PCR of the processing target group arranged on the output table is added to PCR_SUM. And the process ST
Proceeding to 56, it is determined whether there is another unplaced connection group having the same PCR. If there is another such connection group, the process returns to step ST43 and the same process is performed on the group to be processed thereafter. If no such connection group exists, the process proceeds to step ST57 to determine whether or not the processing for all connection groups has been completed. If there is a connection group for which processing has not been completed, the process returns to step ST42 to continue the processing.

As described above, according to the sixth embodiment, when the connection to be processed cannot be located at the ideal location address while the PCR_SUM is equal to or less than half of the total output bandwidth value of the output table, the ideal Since the arrangement is made such that the arrangement is preferentially arranged at odd addresses (or even addresses) within the range of the CDV value centered at the arrangement address, a more uniform arrangement of the connection groups on the output table is realized, and Since collision and congestion are prevented, an effect is obtained that a larger number of connections can be secured for the allowable CDV value.

In the case where an odd address (or an even address) is searched for preferentially, the allowable C value centered on the ideal arrangement address is used.
Since the search is made to search for odd addresses (or even addresses) in the front and rear within the DV value, empty addresses in the range of the allowable CDV value can be efficiently used. The number of connections can be ensured, and even if the collision of ATM cells frequently occurs, the effect that the output band is not reduced unlike the search only for the rear is obtained.

Further, the search is switched from the search at the odd address to the search at the even address on the basis of the point in time when the PCR_SUM becomes half of the total output band value of the output table, so that the search of the empty address can be easily performed. As a result, the output table can be created more efficiently. Further, the same effect as in the fifth embodiment can be obtained in the configuration including the wide band output table.

In the process ST101 of FIG. 72, if PCR_SUM is equal to or less than half of the total output band value, an odd address is searched in the process ST102. However, an even address may be searched. The effect of is obtained.

Embodiment 7 FIG. An apparatus configuration including a shaping circuit and a routing switch for realizing the shaping method according to the seventh embodiment, and a wideband output table used for shaping are shown in FIG.
69 are the same as those shown in FIG. 69, and therefore, the same symbols are used and their detailed description is omitted here.

Next, the operation will be described. FIG. 73 is a flowchart showing a shaping method according to Embodiment 7 of the present invention. 73, the same reference numerals as those in FIG. 71 indicate the same processing in the shaping method, and thus the description thereof will be omitted. The seventh embodiment differs from the sixth embodiment in that step ST44 is omitted.

That is, even if the ideal arrangement address is calculated,
The arrangement address search processing 1 is performed without determining whether or not the arrangement to the ideal arrangement address is possible. Therefore, for one connection group, odd addresses or even addresses are allotted with priority to a plurality of assigned addresses. Note that other operations are the same as those in the sixth embodiment, and a description thereof will be omitted.

As described above, according to the seventh embodiment, substantially the same effects as those of the sixth embodiment can be obtained. The difference is that while the PCR_SUM is equal to or less than half of the total output bandwidth value of the output table, the connection group to be processed is an odd number on the wide output table regardless of whether the ideal arrangement address is an empty address. Addresses (or even addresses) are all given priority,
A very even distribution on the output table is achieved,
The effect is obtained that a larger number of connections can be secured for the allowable CDV value.

[Embodiment 8] An apparatus configuration including a shaping circuit and a routing switch and a wideband output table used for shaping for realizing the shaping method according to the eighth embodiment are shown in FIG.
69 are the same as those shown in FIG. 69, and therefore, the same symbols are used and their detailed description is omitted here.

Next, the operation will be described. FIG. 74 is a flowchart showing a shaping method according to the eighth embodiment of the present invention. In FIG. 74, the same reference numerals as those in FIG. 71 indicate the same processing in the shaping method.
The description is omitted. The eighth embodiment is different from the sixth embodiment in that the arrangement address search processing 1 in the process ST46 is replaced with the arrangement address search processing 2.

FIG. 75 is a flow chart showing the contents of the layout address search processing 2. In this embodiment, for simplicity of explanation, the explanation will be made assuming that the allowable CDV value is 7. In the arrangement address search processing 2, first, the processing ST
At 151, it is determined whether or not PCR_SUM is equal to or less than half of the total output bandwidth value (1.2 Gbps in the eighth embodiment) of the output table. If less than half, processing ST
At 152, 1 is set as the initial value of the loop counter variable i. Then, in a process ST153, it is determined whether or not PCR_SUM is not more than {(output band) / (allowable CDV value + 1)} * i. In this embodiment, since the output band is 1.2 Gbps and the allowable CDV value is 7, the above equation is given as 150 Mbps * i.

In the process ST153, the PCR_SUM
Is 150 Mbps * i or more, the process ST1
In step 54, i is incremented by one, and then in process ST155, it is determined whether or not i is equal to or smaller than an allowable value. This tolerance is
For example, a value calculated as (allowable CDV value + 1) / 2 corresponding to a situation where the value of PCM_SUM is half of the total output band value is used. In the present embodiment, 4 is used as the allowable value. If i is equal to or smaller than the allowable value in the process ST155, the process returns to the process ST153 to continue the process. If i is equal to or more than the allowable value in process ST155,
The process returns to step ST151 to continue the process.

If PCR_SUM is equal to or less than 150 Mbps * i in process ST153, the process ST15 is executed.
Proceed to 6. In the process ST156, when the address on the output table is divided by 8 (allowable CDV value + 1) (predetermined number) near the ideal arrangement address based on the value of i, an odd address having a specific remainder is searched.

For example, it is possible to provide the following search procedure based on the value of i. 1) When i = 1: Search for an address where the remainder when the address is divided by 8 is 5 2) When i = 2: The remainder when the address is divided by 8 is 1,5
3) When i = 3: the remainder when the address is divided by 8 is 1,
4) When i = 4: the remainder when the address is divided by 8 is 1,
Searching for Addresses 3, 5, 7 As described above, empty addresses are limitedly searched in the vicinity of the ideal arrangement address according to the value of i.

When the search based on the value of i is completed,
In process ST157, it is determined whether or not arrangement is possible. If it can be arranged, the arrangement address search processing 2 is terminated in step ST158 as "arrangeable". If arrangement is not possible, the process proceeds to step ST154 to continue the process.

If the value of PCR_SUM is larger than half of the total output band value in the output table in process ST151,
Proceeding to process ST159, the allowable C near the ideal arrangement address
An empty address of an even address is searched in the range within the DV value. Then, it is determined in step ST160 whether or not the arrangement is possible. If the arrangement is possible, the process proceeds to step ST158, and the arrangement address search processing 2 is terminated as "arrangeable".

If arrangement is not possible, processing ST161
Then, an empty address is searched for regardless of the even address or the odd address within the range of the allowable CDV value near the ideal arrangement address.
Then, it is determined whether or not arrangement is possible in process ST162,
If the arrangement is possible, the process proceeds to step ST158 to set the address as "possible", and the arrangement address search process 2 is ended. I do.

As described above, according to the eighth embodiment, substantially the same effects as those of the sixth embodiment can be obtained. The difference is that, based on the occupancy of the output table derived from the PCR_SUM value, the connection group is arranged at a specific address where the remainder when divided by a predetermined number has a specific value. ,
Since the arrangement of the connection groups on the output table is appropriately controlled according to the occupancy of the output table and arranged evenly, collision and congestion of ATM cells are prevented and the allowable CD is prevented.
The effect is obtained that a larger number of connections can be secured for the V value.

In the process ST151 of FIG. 75, if PCR_SUM is equal to or less than half of the total output band value, an odd address is searched in the process ST156. However, an even address may be searched. The effect of is obtained.

[Embodiment 9] An apparatus configuration including a shaping circuit, a routing switch, and the like for realizing the shaping method according to the ninth embodiment, and a wideband output table used for shaping are shown in FIG.
69 are the same as those shown in FIG. 69, and therefore, the same symbols are used and their detailed description is omitted here.

Next, the operation will be described. FIG. 76 is a flowchart showing a shaping method according to Embodiment 9 of the present invention. In FIG. 76, since the same reference numerals as those in FIG. 74 indicate the same processing in the shaping method,
The description is omitted. The ninth embodiment differs from the eighth embodiment in that step ST44 is omitted.

That is, when the ideal arrangement address is calculated,
The arrangement address search processing 2 is performed without determining whether or not the arrangement to the ideal arrangement address is possible. Therefore, for one connection group, odd addresses or even addresses are allotted with priority to a plurality of assigned addresses. Note that other operations are the same as those in the eighth embodiment, and a description thereof will not be repeated.

As described above, according to the ninth embodiment, substantially the same effects as those of the eighth embodiment can be obtained. The difference is that, regardless of whether or not the ideal arrangement address is an empty address, the group to be processed is preferentially arranged at an odd address (or an even address) on the broadband output table and is derived from the PCR_SUM value. Based on the occupancy of the output table, the remainder is arranged at a specific address having a specific value, so that a very uniform arrangement is realized on the output table, so that more residual CDV values are required. The effect that the number of connections can be secured is obtained.

[0265] In the empty address search in the process ST156, the priority in odd addresses is represented by a remainder of 5,
Although the addresses are arranged in the order of 1, 7, and 3, this is selected with the intention of widening the interval between the already allocated addresses even at the odd addresses. The order is not limited to this, and the same effect can be obtained by selecting another effective order.

[0266]

As described above, according to the present invention, the ideal position calculating process includes the integer interval calculating process, the real number interval calculating process, the bias adding process, the interval mixing method changing process, and the group ideal position calculating process. It includes an even distribution process, an intra-group start position uniform distribution process, a matrix coordinate storage process, a start position random distribution process, a common equalization distribution process, and an adjacent placement prohibition process. Processing, arrangement ratio dependent pattern matching processing, neighborhood random search processing,
Providing the whole random search processing and the whole continuous random search processing, if the arrangement processing time is short and it is not possible to arrange at the ideal arrangement position, select an appropriate alternative position and perform an evenly distributed arrangement as a whole. There is an effect that can be.

According to the present invention, by providing the decimal bias adding process in the bias adding process, the bias changing process and the bias value changing pattern storing process in the interval mixing method changing process, the result of the integer conversion can be varied. This is advantageous in that the possibility of rearrangement can be increased when the camera cannot be arranged at the ideal arrangement position.

According to the present invention, in the real number interval calculation processing, the first point is determined by using a fixed-point variable of 32 bits or less.
By obtaining the remainder answer and the second remainder answer, it is possible to shorten the arrangement processing time by processing with only a small number of bits and using only integers.

According to the present invention, a random number initialization process, a random number initialization process, a periodic random search process, a whole random search process, and a whole continuous random search process in an alternative position search process in a start position random distribution process in an ideal position calculation process. Includes processing for updating each variable with a linear function, random number generation cycle control processing, cycle selection processing, and random number function calling processing, thereby increasing the possibility of rearrangement when placement at an ideal placement position is not possible There is an effect that can be.

According to the present invention, in the pattern matching processing in the alternative position search processing, the optimum arrangement position pattern creation processing, the optimum arrangement position pattern storage processing, and the optimum arrangement position pattern reference processing are provided, so that the whole is provided. There is an effect that they can be arranged evenly.

According to the present invention, in the pattern matching processing and the neighborhood random search processing in the alternative position search processing, the index calculation processing, the narrowing / borrow update processing, and the borrow accumulation processing are provided, so that the entirety is evenly distributed. There is an effect that it can be arranged.

According to the present invention, if the group to be processed cannot be arranged at the ideal arrangement position while the total band value already arranged is smaller than the determination value, the CDV value within the ideal arrangement position is the center. Since it was configured to preferentially place at odd addresses (or even addresses) in the range,
The group is evenly arranged on the parent table
There is an effect that TM cell collision and congestion are prevented, and a larger number of groups can be secured for the allowable CDV value.

According to the present invention, while the total allocated band value is smaller than the determination value, regardless of whether or not the ideal allocation position is an empty address, the group to be processed is displayed on the broad base table. Since it is configured so that it is allocated preferentially to all odd addresses (or even addresses),
Because a very even arrangement is realized on the maternal table,
There is an effect that a larger number of groups can be secured for the allowable CDV value.

According to the present invention, since the judgment value to be compared with the total band value is provided as a half value of all the output bands of the matrix table, the arrangement for most odd addresses is completed. Since the search can be switched to the even address search later, an empty address can be easily searched, and the creation of the mother table can be made more efficient.

According to the present invention, when a group to be processed is preferentially arranged at an odd address or an even address within a range within the allowable CDV value, the remainder divided by a predetermined number is processed to an address having a specific numerical value. Since the target groups are arranged, the arrangement of the groups on the parent table is more controlled and the groups are arranged evenly.
There is an effect that the collision and the congestion of the ATM cells are prevented, and a larger number of groups can be secured for the allowable CDV value.

[Brief description of the drawings]

FIG. 1 is a diagram showing a GID table band according to Embodiment 1 of the present invention.

FIG. 2 is a diagram showing a GID table ring according to the first embodiment of the present invention.

FIG. 3 is a diagram illustrating terms according to the first embodiment of the present invention.

FIG. 4 is a conceptual diagram of band allocation according to the first embodiment of the present invention.

FIG. 5 is a diagram showing an example of an input pattern according to the first embodiment of the present invention.

FIG. 6 is a diagram illustrating priorities according to the first embodiment of the present invention.

FIG. 7 is a diagram showing a group hierarchy according to the first embodiment of the present invention.

FIG. 8 is a flowchart showing a GID arrangement process according to the first embodiment of the present invention.

FIG. 9 is a flowchart showing an initial setting according to the first embodiment of the present invention.

FIG. 10 is a flowchart showing the same band group group arrangement processing according to the first embodiment of the present invention.

FIG. 11 is a flowchart showing a group arrangement process according to the first embodiment of the present invention.

FIG. 12 is a flowchart showing individual group element arrangement processing according to the first embodiment of the present invention.

FIG. 13 shows a function cnt according to the first embodiment of the present invention.
It is a flowchart which shows _mix_clc_int.

FIG. 14 is a diagram showing a comparison between CDV and borrowing by the calculation method according to the first embodiment of the present invention.

FIG. 15 shows a function cnt according to the first embodiment of the present invention.
13 is a flowchart showing _mix_real.

FIG. 16 is a diagram illustrating a difference due to a difference in an integer conversion method according to the first embodiment of the present invention.

FIG. 17 is a diagram for explaining integerization and rounding bias according to the first embodiment of the present invention.

FIG. 18 is a diagram showing a concept of matrix coordinates according to the first embodiment of the present invention.

FIG. 19 is a diagram showing an example of matrix coordinates according to the first embodiment of the present invention.

FIG. 20 is a diagram showing an example of matrix coordinates according to the first embodiment of the present invention.

FIG. 21 is a diagram illustrating a process of obtaining matrix coordinates from a GID according to the first embodiment of the present invention.

FIG. 22 shows a function cnt according to the first embodiment of the present invention.
13 is a flowchart illustrating _matrix_coordinate_set.

FIG. 23 is a diagram illustrating an arrangement of one group according to the first embodiment of the present invention.

FIG. 24 is a diagram illustrating an arrangement of two groups in the same band according to the first embodiment of the present invention.

FIG. 25 is a diagram showing matrix coordinates of the same-band two-group arrangement according to the first embodiment of the present invention.

FIG. 26 shows a function cnt according to the first embodiment of the present invention.
It is a flowchart which shows _set_step_inf.

FIG. 27 is a function cnt according to the first embodiment of the present invention;
13 is a flowchart illustrating _set_size_inf.

FIG. 28 is a diagram illustrating steps and sizes according to the first embodiment of the present invention.

FIG. 29 is a diagram showing an example of a size information holding table according to the first embodiment of the present invention.

FIG. 30 is a diagram showing an example of a size arrangement according to the first embodiment of the present invention;

FIG. 31 is a diagram for explaining equal distribution of arrangement start steps according to the first embodiment of the present invention.

FIG. 32 shows a function cnt according to the first embodiment of the present invention.
13 is a flowchart illustrating _set_start_offset.

FIG. 33 shows a function cnt according to the first embodiment of the present invention.
It is a flowchart which shows _set_clc_gid.

FIG. 34 shows a function cnt according to the first embodiment of the present invention.
It is a flowchart which shows _chk_gid_available.

FIG. 35 is a diagram showing a treatment when it cannot be arranged at an ideal position according to the first embodiment of the present invention.

FIG. 36 is a diagram illustrating an ideal position and a periphery according to the first embodiment of the present invention.

FIG. 37 is a diagram illustrating a mechanism of a ROM table and pattern matching according to the first embodiment of the present invention.

FIG. 38: Function cnt according to Embodiment 1 of the present invention
It is a flowchart which shows _chk_cdv_pat_value.

FIG. 39 shows a function cnt according to the first embodiment of the present invention.
It is a flowchart which shows _chk_cdv_pat_R1357.

FIG. 40 is a function cnt according to Embodiment 1 of the present invention;
It is a flowchart which shows _chk_start_rand_all.

FIG. 41 shows a function cnt according to the first embodiment of the present invention.
It is a flowchart which shows _cnk_cdv_rand.

FIG. 42 is a flowchart showing a CDV calculation algorithm according to the first embodiment of the present invention.

FIG. 43 is a diagram illustrating a mechanism of CDV calculation according to the first embodiment of the present invention.

FIG. 44: Permissible CDV according to Embodiment 1 of the present invention
It is a figure explaining an increase of a value.

FIG. 45 shows cnt_s according to the first embodiment of the present invention.
It is a flowchart which shows et_allow_cdv.

FIG. 46 is a diagram showing contents held in a random number table according to the first embodiment of the present invention.

FIG. 47 is a function cnt according to Embodiment 1 of the present invention;
13 is a flowchart showing _rand.

FIG. 48 shows a function cnt according to the first embodiment of the present invention.
12 is a flowchart illustrating _sland.

FIG. 49 is a diagram showing a comparison between a random number and a built-in random number of C according to the first embodiment of the present invention.

FIG. 50 shows a case where the random number according to the first embodiment of the present invention is 11
It is a figure which shows the number which generate | occur | produced exactly once at the time of calling 980 times.

FIG. 51: Function cnt according to Embodiment 1 of the present invention
13 is a flowchart showing _log2.

FIG. 52 is a diagram showing parameter definition according to the second embodiment of the present invention.

FIG. 53 is a flowchart showing an arrangement position calculation method using double precision in the case of multiplication preceding according to Embodiment 2 of the present invention;

FIG. 54 is a flowchart showing an arrangement position calculation method using double precision in the case of division precedence according to Embodiment 2 of the present invention;

FIG. 55 is a diagram for explaining calculation by fixed-point notation according to the third embodiment of the present invention.

FIG. 56 is a flowchart showing an arrangement position calculation method using fixed-point notation in the case of multiplication preceding according to Embodiment 3 of the present invention.

FIG. 57 is a diagram for explaining the required number of bits for fixed-point calculation in the case of multiplication-precedence according to the third embodiment of the present invention.

FIG. 58 is a flowchart showing an arrangement position calculation method using fixed-point notation in the case where an integer is used before division according to Embodiment 3 of the present invention;

FIG. 59 is a flowchart showing an arrangement position calculation method using fixed-point notation when a real number is used in advance of division according to Embodiment 3 of the present invention;

FIG. 60 shows res_r according to Embodiment 4 of the present invention.
It is a flowchart which shows an es method.

FIG. 61 is a diagram illustrating parameters according to the fourth embodiment of the present invention.

FIG. 62 is a diagram showing the relationship between the maximum value / upper limit value of each variable and the band according to the fourth embodiment of the present invention.

FIG. 63 shows res_r according to Embodiment 4 of the present invention;
It is a flowchart which shows an es method.

FIG. 64 is a diagram showing a relationship between a maximum value of a rounding error and round_bias_int according to the fourth embodiment of the present invention.

FIG. 65 is a diagram illustrating a maximum value of res_ans_s when a rounding error is added according to the fourth embodiment of the present invention.

FIG. 66 is a diagram showing evaluation indexes and results according to the fourth embodiment of the present invention.

FIG. 67 shows a double according to Embodiment 4 of the present invention;
It is a figure which shows the number when e and answer of a res_res method differ.

FIG. 68 is a schematic diagram showing a configuration for realizing a shaping method according to Embodiment 5 of the present invention.

FIG. 69 is a diagram showing an example of an output table according to the fifth embodiment of the present invention.

FIG. 70 is a flowchart showing a shaping method according to Embodiment 5 of the present invention.

FIG. 71 is a flowchart showing a shaping method according to Embodiment 6 of the present invention.

FIG. 72 is a flowchart showing the contents of a location address search process 1 according to Embodiment 6 of the present invention;

FIG. 73 is a flowchart showing a shaping method according to Embodiment 7 of the present invention.

FIG. 74 is a flowchart showing a shaping method according to Embodiment 8 of the present invention.

FIG. 75 is a flowchart showing the contents of a location address search process 2 according to Embodiment 8 of the present invention;

FIG. 76 is a flowchart showing a shaping method according to Embodiment 9 of the present invention.

FIG. 77 is a diagram illustrating a conventional shaping method.

FIG. 78 is a flowchart showing a GID arrangement procedure of a conventional shaping method.

[Explanation of symbols]

1 CPU, 2 shaping circuit, 3 switch section,
4 input interface unit, 5 output interface unit, 10 output table (group identifier arrangement table), 11 unit table (slot).

[Procedure amendment]

[Submission date] September 3, 1999 (1999.9.3)

[Procedure amendment 1]

[Document name to be amended] Statement

[Correction target item name] 0179

[Correction method] Change

[Correction contents]

Whole_divide_db = whole_db / divider_db;/* The point that this division does not truncate is the point. * / Whole_divide_db_shift = whole_divide_db; for (p = 1; p <= SHIFT; p ++) ｛whole_divide_db_shiftt  = Whole_divide_db_shift * 2;} / * SHIFT bit shift of real number * / whole_divide_db_shift_int = (unsigned long) whole_division of / * correspond to / first_of * line / derivation of the first expression in / * correspond to / first of this expression; Thereafter, the same processing as in equation (9) is performed using an integer. * / Answer_int = whole_divide_db_shift_int * i; answer_int = answer_int >> SHIFT; (12)

[Procedure amendment 2]

[Document name to be amended] Statement

[Correction target item name]

[Correction method] Change

[Correction contents]

An example of the program is as follows, and its flowchart is shown in FIG. whol_quo = whole / divider; whol_res = whole% divider; res_s = whole_res <<S; res_quo_s = res_s / divider; res_res_divers_res = diversizer
int << S) / round_bias_pitch; / * division without the parentheses of the above equation is performed first want intends * / / * Up to this point, does not depend on i calculated * / res_quo_ans_s = res_quo_s *
i; res_res_ans_s = res_res_s * i
/ Divider; res_ans_s = res_quo_ans_s + r
es_res_ans_s + round_bias_
s; res_ans = res_ans_s >>S; quo_ans = whole_quo * i; answer = quo_ans + res_ans;

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Yasuhito Sasaki 2-3-2 Marunouchi, Chiyoda-ku, Tokyo Sanishi Electric Co., Ltd. (72) Hironori Terauchi 2-3-2 Marunouchi, Chiyoda-ku, Tokyo 3 Inside Ryo Denki Co., Ltd. (72) Inventor Shunsuke Tsutsumi 2-3-1 Otemachi, Chiyoda-ku, Tokyo Nippon Telegraph and Telephone Corporation (72) Inventor Manabu Yoshino 2-3-1 Otemachi, Chiyoda-ku, Tokyo F-term (reference) in Nippon Telegraph and Telephone Corporation 5K030 GA01 HA10 HB29 KA03 LB19 LC02 LE01 MB12

Claims (10)

[Claims] 1. A shaping method for dispersing and arranging a number of elements of one or a plurality of groups, which are determined for each group, in a matrix table composed of a plurality of slots having addresses. Grouping processing for arranging groups together; priority allocation processing for arranging the groups with higher priority; descending order processing for arranging the groups with higher bandwidth; ideally arranged slot positions for each group (hereinafter, slot positions) Is called a position), a group element arranging process for arranging a group element at the calculated ideal arrangement position, and an alternative position search for searching for an alternative slot when the element cannot be arranged at the ideal arrangement position The ideal position calculation process, wherein a predetermined number of the groups are transferred to the matrix table. An integer interval calculation process for obtaining two types of integer arrangement intervals at the time of equal arrangement, and an ideal arrangement position for uniformly arranging a predetermined number of groups in the matrix table by real number calculation and converting it into an integer. A real number interval calculating process, and a bias adding process of adding a bias and then performing an integer calculating operation when converting the value obtained by the real number calculation into an integer in the real number interval calculating process; and the two types of integer arrangement intervals. An interval mixing method changing process of changing the selection order of the group; an ideal position equal distribution process in the group for equally distributing the ideal arrangement position of each group within the same band group within the integer arrangement interval; Intra-group start position equal distribution processing for equally distributing the arrangement start position for each group in the matrix table; real number interval calculation processing; and intra-group ideal position An ideal arrangement position calculated by the equal dispersion processing, a matrix coordinate storage processing for storing the distribution start position calculated by the intra-group start position equal distribution processing as matrix coordinates, and an arrangement start position of the group having a different band, A start position random distribution process in which random distribution is performed in the matrix table; Adjacent placement prohibition processing for prohibiting the placement of elements in adjacent slots, wherein the alternative position search processing observes a slot arrangement situation around the ideal arrangement position,
A pattern matching process for obtaining an optimal arrangement position by comparing with an optimal arrangement position pattern stored in advance, and an arrangement ratio dependent on changing the optimal arrangement position pattern depending on the ratio of arranged slots in the mother table. A pattern matching process, a near random search process for randomly searching for a nearby slot at the ideal placement position, a whole random search process for randomly searching for a placement start position of the group in the matrix table, A continuous random search process for randomly searching for an arrangement start position reference point in the matrix table in a predetermined continuous slot. 2. A bias adding process in an ideal position calculating process in which a bias value to be added is set to a value equal to or greater than zero and less than 1; and a decimal bias adding process in an interval mixing method changing process in the ideal position calculating process. In the bias change process of changing the bias value and resetting the group element, and in the bias change process, the order of the bias value to be changed is determined in advance, and the bias value that takes a value of zero or more and less than 1 is set to the bias value. 2. A shaping method according to claim 1, further comprising a sequence of integer values and a bias value change pattern storing process for storing the total number of types of bias values as integer values by multiplying the total number of types. . 3. In a real number interval calculation process in an ideal position calculation process, a parameter is divided by a divisor using a fixed-point variable of an integer of 32 bits or less to obtain a first quotient and a first remainder. A second processing of shifting the first remainder leftward by a predetermined number of bits (multiplying 2 by a predetermined number of times), and dividing the left-shifted first remainder by the divisor A third process for calculating a second quotient and a second remainder, and a fourth process for multiplying the second quotient by a first integer equal to or larger than 1 and equal to or smaller than the divisor to obtain a first variable. A fifth process of multiplying the second remainder by the first integer and dividing by the divisor to obtain a second variable; and adding the first variable and the second variable; A sixth processing for obtaining a first surplus answer; and a method for converting the first surplus answer to the predetermined number of bits. 7. Shift right (divide by 2 for a predetermined number of bits) and obtain a second surplus answer. Multiply the first quotient by the first integer to obtain a quotient answer. An eighth process, a ninth process for obtaining an answer by adding the second surplus answer and the quotient answer, and a first integer
2. The shaping method according to claim 1, further comprising: a loop process of changing the number from 1 or more to the divisor or less and repeating the fourth process to the ninth process. 4. A cycle of a random number represented by a power of 2 in a random distribution process of a start position in an ideal position calculation process, a neighborhood random search process, a whole random search process, and a whole continuous random search process in an alternative position search process. , A variable that is different for each cycle, and a variable that initializes the variable for each cycle, a process that updates the variable for each cycle with a linear function, and zero and maximum destination when a random number is generated for one cycle A random number generation cycle control process for randomly generating each integer between once or twice, a cycle selection process for selecting a minimum power-of-two cycle that does not exceed the maximum destination, and a power-of-two selected above 2. The shaping method according to claim 1, further comprising a random number function calling process for calling a random number function having a period of: 5. In the pattern matching processing in the alternative position search processing, an optimum arrangement position pattern creation processing performed in advance, an optimal arrangement position pattern storage processing performed in advance, and an optimal arrangement position pattern reference processing performed in pattern matching are performed. If the above-mentioned optimal arrangement position pattern creation processing includes empty positions that are equally suitable, there is a process in which all of them are arrangement candidates. A process of setting the vacant position as an arrangement candidate, and a process of setting a position with the smallest number of arranged continuations as an arrangement candidate when there is no vacant position where the arranged continuation does not occur; The patterns to be stored in the position pattern storage processing are as follows: the arrangement state of the slots around the ideal arrangement position is set as 1 for the arranged slots and 0 for the empty slots. An arrangement pattern that is expressed as a continuous number sequence of 1s and 0s and is converted into a decimal number by regarding this as a binary number sequence, the number of arrangement candidates in the optimal arrangement position pattern, and the arrangement candidate position in the optimal arrangement position pattern is represented by an ideal arrangement. A layout candidate position represented by a relative position with respect to the position, wherein the optimal layout position pattern reference process investigates the layout status of slots around the current ideal layout position and creates a layout status pattern; A process of referring to the pattern stored in the optimal arrangement position pattern storage process; and a process of selecting one of the arrangement candidate positions by using a random number when there are a plurality of arrangement candidates stored in the optimal arrangement position pattern storage process. The shaping method according to claim 1, further comprising: 6. In a pattern matching process and a neighborhood random search process in an alternative position search process, an alternative position actually arranged when an ideal position cannot be arranged is shifted to the right of the mother table from the ideal position (a side with a larger slot address). Is added to the calculated ideal arrangement position address, and the index is obtained by subtracting the substitute position address to obtain an index. Calculation processing, and when the index is positive or zero, the difference in the number of slots when the alternate position is located on the left side (smaller slot address) in the mother table from the ideal position (the difference between the alternate position and the ideal position). Address difference) as an index, and new borrowing as zero; if the index is negative, the narrowing as zero 2. The borrowing and borrowing updating process of setting the new borrowing as an absolute value of the index and a borrowing accumulating process of adding the new borrowing to the accumulated value of the borrowing so far. Shaping method. 7. A shaping method for adjusting a transmission interval of fixed-length cells transmitted based on an output band determined for each connection of one or a plurality of connections, comprising: a connection or a set of several connections; A process of rearranging groups in the order of larger bandwidth values required for a configured connection group (hereinafter collectively referred to as a group), and transmitting fixed-length cells to a group to be processed in the order of bandwidth values. Processing for calculating an ideal transmission interval of a plurality of slots each having an address, and a matrix table for defining the output order of fixed-length cells,
A process of sequentially determining ideal placement positions for all placement positions including a placement start position for a processing target group; a process of determining whether a processing target group can be placed at the ideal placement position; In some cases, the processing of arranging the group to be processed at the ideal arrangement position, and when the arrangement is not possible, when the total value of the arranged bands is smaller than the determination value, the tolerance centered on the ideal arrangement position is allowed. A group to be processed is preferentially arranged at odd addresses (or even addresses) within the range of the CDV value, and when the arranged band total value is larger than the determination value, processing is performed at an arbitrary address within the range of the allowable CDV value. And a process of arranging a target group. 8. A shaping method for adjusting a transmission interval of fixed-length cells transmitted based on an output band defined for each connection of one or a plurality of connections, comprising: a connection or a set of several connections; A process of rearranging groups in the order of larger bandwidth values required for a configured connection group (hereinafter collectively referred to as a group), and transmitting fixed-length cells to a group to be processed in the order of bandwidth values. A process of calculating an ideal transmission interval of a plurality of slots, each of which has an address, on an output table that defines the output order of fixed-length cells,
The process of sequentially determining the ideal arrangement position for all the arrangement positions including the arrangement start position for the processing target group, and when the total band value already arranged is smaller than the determination value,
If the group to be processed is preferentially arranged at odd addresses (or even addresses) within the allowable CDV value centered on the ideal arrangement position, and if the total allocated band value is larger than the determination value, the ideal Allowable CD centering on the placement position
Arranging a group to be processed at an arbitrary address within a range of the V value. 9. The shaping method according to claim 7, wherein the judgment value to be compared with the total band value is given as a half value of all the output bands of the matrix table. 10. When a group to be processed is preferentially arranged at an odd address (or an even address) within a range of allowable CDV values, a target to be processed is assigned to an address whose remainder obtained by dividing by a predetermined number becomes a specific numerical value. 9. The shaping method according to claim 7, wherein groups are arranged.