KR900002649B1 - Load-distribution system of a digital switching network - Google Patents

Abstract

The method comprises a first process (5a) for checking the range of time switch, a second process (5b-d) for selecting maximum nonoccupation group and initializing the number of maximum searched channel, a third process (5e,5f) for checking the presence of nonoccupied channel by receiving data from occupation/nonoccupation map, a fourth process (5h) for executing normal routine by calculating the channel number from map pointer and bit position when the nonoccupied channel is present, and a fifth process (5g-5h) for searching nonoccupied channel by checking through the last group when the non-occupied channel is not present in the third process.

Description

Load-distributed channel discovery method of digital switching network

1 is a conventional flow diagram.

2 is a database configuration diagram according to a conventional search method.

3 is a system diagram according to the present invention.

4 is a database configuration diagram of a load distribution method according to the present invention.

5 is a flow chart according to the present invention.

6 is a subflow diagram of FIG. 5 according to the present invention.

* Explanation of symbols for main parts of the drawings

MUX ₁ -MUX _N : Multiplexer DMUX ₁ -DMUX _N : Demultiplexer

TSW ₁ -TSW ₈ : Time switch SSW: Space switch

TP: Conversion Processor

The present invention relates to a load distributed channel retrieval method of a digital switching network.

In general, the electronic switching center having a TST (T1Me-Space-T1Me) structure of a digital switching switching network (hereinafter referred to as SN) is used to transmit and receive PCM data through an internal channel (Intermediate T1Me Slot). do.

Here, how to allocate internal channels in software determines SN performance.

In the conventional TST structure, the internal channel allocation method of the digital switch switching network searches from the pre-search completion point in the occupied / not occupied map indicating the internal channel usage of the T1Me switch. . If a non-occupied channel is found here, the point where the non-occupied channel is found is stored and used again in the next search.

That is, the time required to search one non-occupied channel increases in proportion to the maximum number of search channels. In order to solve this problem, reducing the maximum number of search channels to reduce the search time increases the blocking probability of the SN, which causes a large constraint on the performance of the SN.

Referring to FIG. 1, the conventional internal channel search method is described in detail using a database as shown in FIG. 2. Referring to FIG. 1, the transmission time of (2a) is shown in FIG. It stores the previous search point of the position (Incoming T-SW), and the map which operates the semantic / non-occupied state of the channel for each transmission time switch of (2b) for each bit, and (2c) It consists of a map of occupied / non-occupied status of each incoming time switch (Outgoing T-SW). The size of these data is determined by the capacity of the SN, that is, it varies depending on the number of timeslots of one T-SW and how the entire SN is composed of several T-SWs. In the SN controller, when the transmission and destination T-SW to be connected are determined, the point of completion of the time switch pre-search completion corresponding to the transmitting T-SW number is taken and the channel search point is combined by combining the transmission and destination T-SW numbers. Decide Then, the channel occupancy / non-occupancy data of the transmission time switch map 2b and the reception time switch map 2c are taken from the search point, and only the channel position where the transmission and destination channels are in the non-occupancy state at the same time is examined. The internal channel number is created by combining the channel location with the non-occupied Search Pointer at the search point.

Since the above method sequentially searches by increasing the search point, the previous channel occupancy state is not known at all.

Therefore, since the loop must be continuously looped until an unoccupied channel is found, the execution time is limited in an exchange that needs to perform processing in real time.

This led to the drawback of not being able to allocate internal channels by examining all available channels.

Therefore, to allocate one internal channel, there is a limit on the number of search channels within the limit allowed by the exchange's execution time. Since the limit of the number of search channels is a factor to increase the blocking probability of SN, this also becomes a factor of exchange performance limitation.

As described above, the conventional channel retrieval method does not solve the conflicting problem between the blocking probability of the SN and the execution time of the exchange, and makes a compromise at an appropriate level. If a non-occupied channel is not found in the channel, it is regarded as SN channel congestion, and thus there is a problem in that a subscriber's call line cannot be connected.

Therefore, an object of the present invention is to load the type to find the search starting point by referring to the group before the search by placing a counter to record the occupied or non-occupied state of the channel by group by grouping the channel range of the digital switch TST structure to be searched in the electronic switchboard An algorithm provides a method for reducing blocking probability and searching time.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

3 is a system diagram according to the present invention, MUX _1- MUX _N is a multiplex TU, TSW _1- TSW ₈ is a time switch, SSW is a space switch, DMUX _1- DMUX _N is a demultiplexer, TP is a conversion processor ( Translation Processor), as shown above, is a TST system and is composed of a multiplexer (MUX), a time switch (TSW), a space switch (SSW), and a demultiplexer (DMX).

Multiplexer (MUX) multiplexes 32 serial 2.048 Mbps PCM buses (32 × 32 = 1.024 timeslots) into one byte (8 bit) parallel 8,192 Mbps PCM buses (8,192 / 8 = 1.024 timeslots) Input to (Tw ₁ -Tw ₄ ), the demultiplexer (DMX) performs the reverse function of the multiplexer (MUX). The space switch (SSW) spatially switches the PCM bus between four input / output time switches (Tw _5- Tw ₈ ) for each time slot, resulting in a switching between 4,096 timeslots (ie, 4,096).

The TST switch unit 10 has a dual structure in order to increase the reliability, and the conversion processor Tp for controlling the same is connected to the subscriber matching module (S1M), the relay line matching module (T1M), and the remote through the dualized T-bus. Communicate with the exchange unit (RSS) matching module (R1M) to form a call.

4 is a configuration diagram of a database of a load-distributed channel search algorithm according to the present invention, wherein 4a is a previous search pointer map of a group Go-Gi of time switches TSW _{1 to} TSW ₈ , and 4b is a time Transmit / no-occupation channel number map of the group (Go-Gi) of positions (TSW ₁ -TSW ₈ ), (4c) is a time switch (TSW ₁ -TSW _N ) transmission occupancy / non-occupancy map, and (4d) is a time The reception occupancy / non-occupancy map of the positions TSW _1- TSW _N.

5 is a flowchart in accordance with the present invention, and FIG. 6 is a sub-flow chart of maximum unoccupied group selection in FIG. 5 in accordance with the present invention.

Therefore, a specific embodiment of the present invention will be described in detail with reference to FIGS. 3-6. A 64KBPS-converted voice signal has a unique time slot number on a multiplexed line and each multiplexer (MUX _1- When input to MUX _N ), SN transmits (T1ME SLOT INTERCHANGE) voice signal to the other party by replacing PCM data in unique timeslot number of two parties on multiplexed line.

The above functions are performed by the time switches TSW _{1 to} TSW ₈ , and the number of time slots processed by one time switch is limited by the access time and capacity of the memory used by the time switches TSW _{1 to} TSW ₈ . SN is composed of several time switches (TSW), and the connection between them is performed by the space switch (SSW). That is, the space switch (SSW) functions to connect the PCM data between the time switches (TSW) in time slot units. For the spatial connection function of the space switch SSW, an internal T1Me slot of the SN is designated by the control of the conversion processor TP. That is, the conversion processor TP examines the range (4 × 4, 16 × 16, 32 × 32) of the transmission and destination time switches Tw _1- Tw ₈ in the process (5a), and then applies the corresponding switch in the process (5b). Check if the standard is correct and return to the error routine when an error occurs in the standard and select the maximum non-occupied group on the database of Fig. 4 in the absence of an error (5c). In FIG. 6 (6a), a pointer register is set by calculating a pointer of a non-occupied channel counter of transmission and reception, and in step 6b, a group processing non-occupancy channel of transmission and reception is summed and set, and (6c) The maximum number of comparisons is determined according to the number of non-occupied channel groups in 1 time switch.In the step (6d), check the loop counter corresponding to the number of groupings in the time switch. Combination and Prescan Pointer Calculate the search start pointer of the grant T _X BI, R _X BU.

If the register 1 is large when the loop counter is not "0" in step (6d), the pointer of register 2 is increased in step (6f) to calculate and set a new register 2 value. Calculate and set the new register 2 value by incrementing the pointer. When the value of register 1 is not greater than the value of register 2 in step (6f), move the value of register 2 to register 1 and repeat step (6h). Perform.

Subsequently, the search number of the non-occupied maximum channel selected in step (5d) is initialized so that there is no error in the data to be read next. In step 5e, data of the transmission / reception occupancy occupancy maps 4c and 4d are transmitted and compared with the maximum channel occupancy. Subsequently, when a non-occupied channel exists in step 4f (5h), the channel number is calculated from the fourth-degree database map pointer and bit position in step 4h to perform a normal function. If not, it checks whether the maximum number of channels corresponding to all channels is checked, and enters the error loop as the channel congestion at the maximum number of channels in step (5i). If it is not the maximum channel, it checks whether it is the last pointer in the group in (5k), and if it is the last point in (5l), it initializes the search point in one group and designates the next channel search point when it is not the last point. That is, the transmission time switches TSW _{1 to} TSW ₄ receive the input time slots, perform time slot exchanges with the internal T1Me slot under the control of the conversion processor TP, and transfer the outputs to the space switches SSW. The space switch (SSW) connects the internal channel data to each incoming time switch (TSW _5- TSW ₈ ), and the incoming time switch (TSW ₅ -TSW ₈ ) accepts the internal channel data and converts the processor (TP). ) to perform a time slot interchange to the destination channel (Outgoing T1Me slot) under the control of the de-peulresyeo (DMUX ₁ -DMUX _N) to send multiple flexure di (DMUX ₁ -DMUX _N) in the jikryeok 8.192Mbps byte parallel data 2.048 Demultiplexed at Mbps and sent to each respective line interface.

The operation of the SN is performed in an orderly manner under the control of the conversion processor TP. The function of the conversion processor TP is between the transmission time switches TSW _{1 to} TSW ₄ and the incoming time switches TSW _{5 to} TSW ₈ . It performs control of the internal channel decision of Intermediate T1Me solt and the space switch (SSW) matrix (4x4 in the case of electronic switch). 16x16 and 32x32.

In order to use the load-distributed channel search algorithm, the same database configuration as the fourth level is required. Load-distributed channel search software distributes the channel's occupied state evenly in one time switch (TSW) rather than in one place, allocating an internal channel (Inetermediate T1Me slot). In other words, the software operates a counter that logically groups the time slots within a time switch (TSW) and records the channel occupancy status for each group. Therefore, since the channel search is started in the group with the most non-occupied channels among the groups in the time switch (TSW), the channel can be allocated without blocking without being aware of all channel occupancy states.

A counter database that records the occupied state of a channel is classified into a sender and a receiver, and their data structure is the same.

Suppose you grouped the total timeslots of one time switch (TSW) into i (this I should be adjusted to the capacity of the T-SW and the performance of the conversion processor (TP)) and the number of channels in one group is (1 TSW T1Me). slot) 1i. For example, if a time switch capacity is 1024T.S · i = 8, the number of channels in one group is 128CH. The number of time switch to be connected is ø2, and the number of unoccupied channels per group of the corresponding time switch

GTø (i) = (36, 47, 28, 80, 30, 50, 70, 62) i = ø1, 2, 3, 4, 5, 6, 7

GRø (i) = (45, 26, 38, 45, 86, 60, 32, 40) i = ø1, 2, 3, 4, 5, 6, 7, and select the group with the most unoccupied channels. The sum of the unoccupied channels of transmission and reception for each group is calculated.

GRø (i) = (81, 73, 66, 125, 125, 110, 102, 102) i = Find the largest of the sum of the non-occupied channels above 1, 2, 3, 4, 5, 6, 7. If the same number of groups appear, the smaller group is chosen first.

In this way, after selecting a group to start searching for a channel, the database (4c) and (4d) pointers take the previous search pointers for each group from (4d) in step (5e) to determine the search starting point. Since the channel search is performed only within the selected maximum non-occupied channel group, the maximum number of search channels is (number of LT SWT1Me slots) / i.

The biggest advantage of the load-distributed channel search algorithm is that the allocation of channels within a time switch (TSW) is distributed and the channel is allocated in consideration of the total channel occupation status (considering the channel occupancy status of the transmitting and receiving TSW). The blocking probability of is significantly reduced compared with the conventional method, and the search time of the non-occupied channel is the highest, so that the execution time of the conversion controller TP is much shorter than that of the conventional method. That is, non-occupied channels can be easily allocated.

These advantages improve the performance of the digital switching network, which is compared with the blocking probability and execution time in the case of the switching network of the electronic exchange as follows.

First, in order to find the blocking probability, the SN blocking probability of the electronic exchange is calculated using the linear graph method of C, Y, and LEE.

a: Channel occupancy, m: 1 The T1Me slot number (1) of TSW is a blocking probability only considering hardware. The channel search algorithm searches only up to 128 channels and considers congestion if there are no unoccupied channels, so the equation (1) changes as follows.

P = (2a-a ² ). … … … … … … … … … … … … … … … … … (2)

Equation (2) is the unilateral blocking probability, so the bidirectional blocking probability is

P = 1- [1- (2a-a ² ) ¹²⁰ ] ² . … … … … … … … … … … … (3)

Since the load-distributed channel search algorithm searches for non-occupied channels in consideration of the overall channel occupancy state, the result is almost the same as searching all 1024 channels. Therefore, the equation (3) in the case of introducing this algorithm changes as follows.

P = 1- [1- (2a-a ² ) ¹⁰²⁴ ] ² . … … … … … … … … … … … (4)

By applying equations (3) and (4), the blocking probability according to channel occupancy is shown in Table 1 below.

TABLE 1

When calculating the SN direction using the above blocking probability, if the conventional algorithm is used, the channel occupancy is 0.76 when P = 0.001, the lowest criterion of loss probability recommended in CC1TT GAS Chapter VI, and SN DYDFD becomes 1556.5 erlang. In case of using a distributed algorithm, the channel occupancy is 0.9154 and the SN capacity is 1871.8 erlang (0.914 × 2048 TS) when P = 0.001, which brings about 20% of the SN capacity.

On the other hand, in terms of execution time, accurate calculation is possible only by simulation (S1Mulation).

In order for the conventional algorithm to maintain the same level of blocking probability as the load-balancing algorithm, the electronic exchange must search all 1024 channels. On the other hand, assuming that the grouping number is 4 when using the method of the present invention, the processor clock used for program coding is 6 MHz, and comparing the execution time of each program shows that the maximum one channel search time is 1.08 msec. 0.43 msec was used for this method. 0.43msec includes the time to find the maximum non-occupied group, and the program structure is almost identical between the two algorithms.

As described above, in the electronic switchboard having the TST structure of the digital switching network, there is a counter that records the occupied / non-occupied status by grouping the channel range to be searched and loads the channel search method to find the search starting point by referring to it before starting the search By using the distributed type, the blocking probability and the high color time can be reduced, thereby increasing the call processing capacity (Call / Hour) per unit time of the entire system.

Claims (1)

In an internal channel allocation method of a digital switch network of a TST structure, a first process (5a) of examining a range of transmission and destination time switches, and a second process of selecting a maximum non-occupancy group and initializing the maximum number of channel searches ( 5b-5d), the third process (5e, 5f) and the third process (5e, 5f) to check the presence of the non-occupied channel by comparing the data from the incoming and outgoing non-occupancy / occupancy map, there exists a non-occupied channel When the channel number is calculated from the map pointer and the bit position, a normal routine is performed. If there is no unoccupied channel in the fourth process (5h) and the third process (5e, 5f), the non-occupied channel is checked until the end of each group. And a fifth process of finding (5g-5n).

Images