JP3870218B2 - Cell flow rate control method and cell exchange system using the same - Google Patents

Description

The present invention is a UPC (Monitoring and Controlling Appropriate Cell Inflow) in UNI (User-to-Network Interface) and NNI (Network-to-Network Interface) of a cell switching system such as an ATM (Asynchronous Transfer Mode) switch. Usage
The present invention relates to a cell flow rate control method having a (Parameter Control) function and a cell exchange system using the same.

In particular, the present invention relates to a cell flow rate control method in which a plurality of lines are multiplexed and a UPC function is performed by one device for cell flows flowing through the multiplexed lines, and a cell switching system using the same.

A cell switching system such as an ATM switch is equipped with a UPC function in a subscriber interface connected to a subscriber terminal through a transmission path. The UPC function is
In general, it is prepared for the subscriber line.

However, even with a relatively low-speed interface, the cell flow control device (
It is uneconomical to prepare a UPC device). Therefore, it is conceivable to multiplex a plurality of the subscriber interfaces and use one UPC device.

FIG. 12 is a block diagram illustrating a configuration example of the ATM switching system. An ATM switch unit 1, an input line corresponding unit 2, and an output line corresponding unit 3 are provided. The ATM switch unit 1 is a self-routing switch such as a Banyan switch, and performs routing to a destination output transmission path based on information in a cell header.

The input line corresponding unit 2 terminates a transmission line (physical line) connected to the subscriber SUB and monitors the flow rate of the cell. The output line corresponding unit 3 has an interface function with the output transmission path, and includes a buffer circuit 30, a maintenance operation management unit 31, and the like.

When considering the use of one UPC device 20 by multiplexing a plurality of physical lines as described above, the input line corresponding unit 2 outputs the plurality of subscriber interface circuits I / F as line multiplexing devices. 21 is input. The cell flow of the signal multiplexed by the line multiplexer 21 is measured by the UPC device 20. Input line support unit 2
Also, a maintenance operation management unit 22 is provided.

The cell that has passed through the UPC device 20 converts the virtual channel identifier (VCI) used on the transmission path into the VCI used by the ATM switch 1 by the header converter 22. According to the converted VCI, the path of the ATM switch 1 is selected and the exchange function is performed.

FIG. 13 assumes a case where two physical lines a and b (subscriber lines) are multiplexed as an example of measuring the multiplexed output cell flow of the line multiplexer 21 with one UPC apparatus 20. FIG. FIG. 14 is a time chart corresponding to FIG.

In FIG. 14, a and b represent cell columns corresponding to the subscriber lines a and b, respectively. 14 is the output of the line multiplexer 21. The cell flows flowing through the subscriber lines a and b are independent of each other and their speed is not common. Therefore, when the timings coincide as in the cells a3 and b4, the line multiplexer 21
It is necessary to delay the timing of any cell.

In the example of FIG. 14, the timing of the cell a3 is delayed. The UPC device 20 grasps the cell train for each subscriber line and calculates the cell flow rate. Therefore, in the cell string corresponding to the subscriber line a separated from the output of the line multiplexer 21, the cell a
The interval T1 between 2 and a3 increases, and the interval T2 between cells a3 and a4 decreases.

For this reason, the arrival of the cell fluctuates at the input point of the UPC device 20. This limits the accuracy of UPC.

Furthermore, in addition to the problem caused by the difference in the cell flow speed flowing through each subscriber line, the UPC accuracy is limited when the sum of the speed after multiplexing and the speed of each interface circuit I / F does not reach an appropriate ratio. There is a problem of being.

FIG. 15 is a diagram for explaining this. For reasons of system configuration, the difference between the speed after multiplexing and the speed of the interface circuit I / F is not an appropriate multiple of the physical line speed. Must be inserted. In FIG. 15, the hatched area (between cells b1 and a2, and between cells b3 and a4) of the multiplexed cell row M is an invalid cell.

As shown in FIG. 15, the inserted invalid cell causes fluctuations in the cell line of the multiplexed a line.

Furthermore, when a plurality of lines are multiplexed and input to one UPC device 20 as another problem, the UPC function must be performed at a speed higher than the sum of the transmission speeds of the respective lines.

However, the UPC device 20 requires floating point arithmetic and the like in order to accurately calculate over a wide range from a low rate to a high rate. Such floating point arithmetic can reduce the capacity of a memory or the like, but requires a lot of calculation time for decimal point arithmetic. For this reason, it becomes a factor which gives a limit to apparatus operation.

Therefore, an object of the present invention is to multiplex a plurality of lines and to make one UPC device 20.
An object of the present invention is to provide a UPU (cell flow rate monitoring) device that solves the above-mentioned problems when attempting to fulfill the UPC function.

Further problems of the present invention will become apparent from the embodiments with reference to the drawings described below.

The first aspect of the cell flow rate control method and the cell switching system using the cell flow rate control method for achieving the object of the present invention is to receive cells sent from a plurality of lines and to control the cell path. And has the following configuration.

First, the plurality of lines are multiplexed. Next, only the time counter corresponding to the line to which the incoming cell belongs is stepped out of the plurality of time counters provided for the line to be multiplexed. And in correspondence with the count value of this advanced time counter,
Control cell flow.

  With the above configuration, the time management of the cell flow rate for each physical line can be performed.

As another configuration, when the physical line to which the cell belongs is not known from the cell identifier, an identifier for identifying the line to which the incoming cell belongs is attached in the multiplexing of the plurality of lines, and then the identifier The time counter corresponding to the line to be identified is incremented.

Further, in the above, when the plurality of lines are multiplexed, a speed adjustment cell that does not belong to any line is inserted, and depending on the speed adjustment cell, no time counter is incremented. As a result, the difference in speed before and after the multiplexing of the plurality of lines can be absorbed.

Furthermore, as another aspect of the present invention, it is provided for the line to be multiplexed.
Each of the plurality of time counters has a counter value shifted from each other for each of the plurality of connections belonging to the corresponding line. Then, counter values shifted from each other for each of the plurality of connections overflow at timing shifted from each other. Thereby, a common counter can be used for a plurality of connections.

Furthermore, as a specific example, a connection that requires the overflow processing,
The line number to which it belongs is compared and stored in the table. When this table is searched by the counter value of the corresponding time counter and the line of the connection to be subjected to the overflow process matches the line to which the connection requiring the overflow process stored in the table belongs The overflow processing is executed for the time counter corresponding to the line to which the connection belongs.

Further, as a specific example, the connection requiring the overflow processing and the line number to which the connection belongs are compared in the table to be stored, and the connection requiring the overflow processing is indicated by a unique ideal arrival time for the connection. It is.

An overflow process is performed by subtracting the counter value of the corresponding time counter from the ideal arrival time.

Further, as a specific example, parameters serving as a reference for controlling the cell flow rate for connections belonging to each of a plurality of lines are stored in a table.

This parameter is recorded in a fixed decimal point, and an exponent value at the decimal point position of the parameter is recorded, and when calculating the parameter and the count value of the time counter, the count value of the time counter is calculated by the exponent value. Move digit and calculate.

Similarly, when the overflow process is performed, the ideal arrival time is represented by an exponent value at a fixed point and a decimal point position. When performing overflow processing by subtracting a specific value related to (maximum counter value + 1) of the corresponding time counter from the ideal arrival time, the specific value is moved by an exponent value digit of the decimal point position. , By subtracting from the ideal arrival time.

Such digit shift avoids the necessity of adjusting the decimal point position each time the UPC parameter is read from the memory.

Embodiments of the present invention will be described below with reference to the drawings. In the drawings, the same or similar elements are described with the same reference numerals or reference symbols.

Before describing the embodiments of the present invention, the features of the UPC device according to the present invention will be described.
A description will be made sequentially.

[Time management for each physical line]
1 and 2 are conceptual diagrams for explaining a first characteristic point of the UPC device of the present invention. When the subscriber line is multiplexed by the line multiplexer 21 as shown in FIG. 12, the problem due to the difference in transmission rate for each physical line is solved.

That is, when the transmission rate for each physical line is different, the time axis to be managed corresponding to each multiplexed line is also different. Therefore, in the present invention, time counters for physical lines are prepared.

In the example of FIG. 1, the UPC device 20 includes a time counter 200 and a selector 201. In the time counter 200, four counters c1 to c4 are prepared corresponding to, for example, four physical lines to be multiplexed. Each time a cell of the corresponding physical line arrives, the selector 201 is configured to increment the time counter prepared for the corresponding physical line among the time counters c1 to c4 of the counter 200.

That is, in FIG. 1, the selector 1 selects the physical line to which the cell belongs from the header portion 11 of the cell 10 output from the line multiplexer 21 (not shown) in FIG. Judgment is made using the Virtual Path Identifier. The selector 201 selects the time counter 20 for the corresponding line from the VPI value.
0 counters c1 to c4 are selected, and the selected counter is incremented by +1 (
Increment).

FIG. 2 shows a time chart showing such a situation. FIG. 2A shows a cell flow in which the output of the interface circuit I / F is multiplexed by the line multiplexer 21 (see FIG. 12). The number given to each cell indicates (VPI-VCI), for example, 3
When -2 is taken as an example, connection number 2 of line number 3 is shown.

Therefore, the count values of the time counters c1 to c4 for each line of the time counter 200 for the cell flow of FIG. 2A are as shown in FIGS. 2B to 2E. Also,
FIG. 2F shows an arrival pattern of a cell corresponding to connection 1 of line 2. In this way, by looking at the value of the time counter corresponding to the line of the counter 200,
The flow rate of the cell can be grasped for each connection belonging to each line number.

[Notification of physical line number]
FIG. 3 is a conceptual diagram illustrating another feature of the present invention. In this feature, in FIG. 1, when the physical line to which the cell belongs is not known from the cell identifier,
The physical line number is newly notified from the line multiplexer 21.

In FIG. 3, when a cell is multiplexed by the line multiplexer 21, a physical line number is added to the cell. Cells CE1, CE2, and CE3 have physical lines a, b,
cell sent by c. Therefore, as shown in FIG.
Cell numbers CE, CE2, and CE3 output from the line numbers a, b, and c, respectively.
Is attached to the header.

The cells with the line numbers a, b, and c are detected by the selector 201 shown in FIG. 1, and the corresponding time counter of the counter 200 is incremented by +1 each time. In such a feature of FIG. 3, an arbitrary VPI / VCI is connected to an arbitrary physical line,
Can be assigned.

[Adjustment of total physical line speed]
Here, in addition to the cell from the physical line, there may be a case where an empty cell prepared for speed adjustment is input to the line multiplexer 21. That is, the cell does not belong to any line. In such a case, as still another feature of the present invention, it is clearly indicated that the cell does not belong to any line.

As shown in FIG. 4, when the effective cells of lines a, b, and c are multiplexed, invalid cell CEX is inserted in line multiplexer 21. Accordingly, as shown in the figure, from the line multiplexer 21, a cell CE1 corresponding to the multiplexed lines a, b, c is obtained.
, CE2 and CE3, an invalid cell CEX for speed adjustment (an explicit x indicating that it is an invalid cell is added to the header portion) is output.

Next, as in FIG. 1, the selector 201 recognizes the corresponding line number of each cell. At this time, the invalid cell CEX is not recognized because there is no corresponding line number.

As a result, the counter corresponding to the time counter 200 is incremented by +1 corresponding to the cell whose line number is recognized. On the other hand, the invalid cell CEX does not affect the progress of any counter.

[Time counter overflow processing]
The UPC function obtains an actual cell flow rate with respect to a previously reported cell flow rate, performs a process such as discarding the cells that exceed the reported cell flow rate, and enables equal operation of the line.

In order to execute this function, the UPC device 20 uses a counter 200 having time counters corresponding to the number of multiple lines as shown in the above feature in order to obtain the actual cell flow rate.

Here, when attention is paid to the respective time counters corresponding to the number of multiplexed lines of the counter 200, the count is incremented for each arrival of the cell of the corresponding physical line.

Each of the plurality of time counters is realized as hardware or software having a predetermined bit length. The count length is not infinite. Furthermore, a plurality of connections belong to one physical line. Therefore, in order to assume the cell inflow amount for each connection, the difference (cell inflow interval) between the previous cell inflow time and the current cell inflow time is calculated. At this time, as described above, the count range of the time counter is finite, and the output of the same count value is repeated at a predetermined cycle.

Therefore, an overflow flag indicating how many times the time counter has overflowed (ie, returned from the maximum count value to 0) between the previous cell inflow time and the current cell inflow time needs to be stored for each connection. is there.

However, as described above, different connections belonging to the same subscriber line are counted in the same time counter of the counter 200. Therefore, when the time counter overflows, if a cell for a connection belonging to the same subscriber line that has not been updated, the inflow interval for that connection is calculated based on the incorrect overflow flag. Will be.

In order to solve such a problem, even when a line is multiplexed according to the present invention and the UPC function is executed by one UPC device, the value of the time counter and the connection identifier (for example, as previously proposed by the present inventors) are proposed. Technology that uses the difference from VPI / VCI as a time counter for the connection (Japanese Patent Application No. 8-30715)
Can be adopted.

In such a technique, a time having a unique offset is defined for each connection, and only the connection can be overflowed when the time returns to 0 (overflow).

That is, for each connection, a time counter value (this is defined as a local time counter value) deviated by a corresponding magnitude for each connection from a reference time counter value (this is defined as a global time counter value). Is used.

As a result, when one local time counter value overflows, the number of connections that need to update the overflow flag is only one. The overflow flag update process for one connection can be completed in a time shorter than the inflow time for one cell.

As a specific example, as shown in FIG. 5, the value of the cell identification number ID uniquely determined from the connection identifier VPI (virtual path identifier) / VCI (virtual channel identifier) of one connection is subtracted from the global time counter value tc. The counter value obtained in this way is defined as the local time counter value tcd corresponding to the connection.

In FIG. 5, the local time counter value tcd overflows at time t1.
That is, when the global time counter value tc becomes equal to the value ID, the local time counter value tcd is overflowed.

The overflow process is performed as follows. That is, TAT (Theoretical
The time corresponding to the length Clen of the time counter is subtracted from the next ideal arrival time indicated by Arrival Time (ideal arrival time), and the obtained value TAT ′ is used as the next ideal arrival time to determine the local time counter value tcd. Continue to advance.

Here, the time corresponding to the length Clen of the time counter can be defined as a specific value related to (maximum counter value + 1) of the time counter. Therefore, the overflow process is performed by subtracting a specific value related to (counter maximum value + 1) of the time counter corresponding to the connection from the ideal arrival time.

In FIG. 5, only the connection identifier ID of the physical line a is shown as the local time counter value tcd. Since the value of the identifier is different for each connection, a time with a unique offset is defined for each connection included in one physical line, and only the relevant connection is overflowed when the time returns to 0 (overflow). can do.

Here, in the configuration in which a plurality of lines are multiplexed and a plurality of time counters corresponding to the physical lines are used, the technique of the previous application is used as it is because there are a plurality of physical lines. I can't. Therefore, in the present invention using a plurality of time counters, the following processing is performed with reference to FIG.

In the present invention, the UPC device 20 has a TAT memory 62. This TA
In the T memory 62, TAT, that is, T set in advance corresponding to each connection.
The physical line number to which the AT (ideal arrival time) belongs is stored in association with the TAT.

Further, similarly to the explanation of the principle of FIG. 1 which is the first feature of the present invention, first, the line number 60 is extracted from the header 11 of the inputted cell 10, and the selector 201 executes the step (
The corresponding time counter to be added (+1 is added) is selected (in the example of the figure, the time counter a is selected with the line number 60 of the input cell 10 corresponding to the line a).

Then, the TAT memory 62 is accessed using the counter value 61 after the addition as an indicator indicating the TAT of the connection to be overflow processed. The TAT indicated here is not directly related to the input cell.

Therefore, the line number stored in the TAT memory is read in correspondence with the TAT that matches the counter value 61 after the addition. Furthermore, when the line number to be read is the same as the line number to which the input cell belongs, it means that the time counter of the connection of the input cell has reached the maximum value, and therefore the overflow processing target Is done.

For example, in FIG. 6, a connection identifier (ID) cell belonging to the physical line a is input, and the time associated with the ID connection at that time is as shown in Table 1.

〓
表１において、「Clen」は、時刻カウンタの長さである。カウンタａの値が０
であっても、コネクション識別番号（ＩＤ）によって、時刻をずらせている。即
ち、図６において、加算後のカウント値６１（グローバル時刻カウンタ値）から
コネクション識別子（ＩＤ）６３が減算回路６４で減算され、これが当該コネク
ションに対する時刻（ローカル時刻カウンタ値）となり、ＵＰＣ処理６５が行わ
れる。この様に、オーバフロー処理が必要なコネクションも１つだけである。

〓
In Table 1, “Clen” is the length of the time counter. The value of counter a is 0
Even so, the time is shifted by the connection identification number (ID). That is, in FIG. 6, the connection identifier (ID) 63 is subtracted from the count value 61 (global time counter value) after the addition by the subtraction circuit 64, and this becomes the time (local time counter value) for the connection, and the UPC process 65 is performed. Done. In this way, only one connection requires overflow processing.

The subtracting circuit 64 inverts the sign of each bit of the connection identifier 63 and adds +1 to this bit number (the 2's complement to the connection identifier 63) to the global time that is the count value 61 after the addition. The subtraction process is executed by executing an operation for adding the count value.

Here, the connection identification number (ID) is unique to the connection. Therefore, as shown in FIG. 6, the connection identifier 63 is subtracted from the added count value 61 (global time counter value) by the subtraction circuit 64, and the result is used as the time (local time counter value) for the connection. In addition, it is possible to configure as shown in FIG.

The global time counter value 61 and the connection identifier 63 in the case of FIG.
Table 2 shows the relationship.

即ち、図７では、加算後のカウント値６１（グローバル時刻カウンタ値）にコ
ネクション識別子６３を加算回路６６で加算するようにしている。この場合も、
コネクション毎に時刻カウンタとして使用する時刻が異なり、オーバフロー処理
が必要なコネクションは１つだけである。

That is, in FIG. 7, the connection identifier 63 is added to the count value 61 (global time counter value) after the addition by the adding circuit 66. Again,
The time used as the time counter is different for each connection, and only one connection requires overflow processing.

If the connection with the identifier ID = 2 in Table 1 belongs to the physical line b, it is shown as Tables 3 and 4.

表３は、物理回線ａについて、表４は、物理回線ｂについて、示している。表
３の物理回線ａには、識別番号ＩＤ＝２は、存在しない。したがって、物理回線
ａに対応する時刻カウンタの計数値がａ＝２であっても、オーバフロー処理は、
行われない。

Table 3 shows the physical line a, and Table 4 shows the physical line b. The identification number ID = 2 does not exist in the physical line a in Table 3. Therefore, even if the count value of the time counter corresponding to the physical line a is a = 2, the overflow process is
Not done.

On the other hand, when the count value of the time counter b corresponding to the physical line b is 2, since there is a connection belonging to the physical line b, an overflow process is required.

As described above, in the TAT memory shown in FIGS. 6 and 7, which line (that is, time counter) the TAT (that is, connection) belongs to, TAT (connection).
Remember in relation to. As a result, overflow processing is executed when the cell belongs to the same physical line as the input cell.

[UPC processing by semi-fixed point arithmetic]
In FIG. 6 or FIG. 7, when the connection-specific time 65 is obtained, the UPC
Processing is performed. Here, the UPC device employs a floating point arithmetic method in order to process connections with a wide range of characteristics from high-speed connections of 150 Mbps or higher to low-speed connections such as 64 Kbps.

However, when the floating-point arithmetic method is adopted, the CPU load required for the calculation becomes large. Therefore, according to the present invention, when the connection-specific time 65 is obtained,
As an optimal UPC process, a UPC process by a semi-fixed point arithmetic method is performed.

The feature of the present invention further enables a UPC apparatus having a small circuit scale without lowering the processing accuracy of the floating-point arithmetic for UPC processing.

As an example according to the present invention, FIG.
This is a flow for performing UPU processing in (Virtual Scheduling Argolithm).

Conventionally, as a cell parameter for each cell, TAT (Theoretiacal Arrival
Time), I (Increment Parameter), and L (Limited Parameter) are stored in a floating format and used for computation. Here, I (Increment Parameter) is an added value expressing a cell inflow interval defined in advance.

Also, L (Limited Parameter) is a problem indicating how early the cell arrives faster than TAT, and is a value indicating an allowable value in this case.

In the present invention, these parameters are all converted into fixed-point numbers, and instead, the positions of the decimal points of these parameters are typically stored in the memory 80 as a parameter called exponent value exp. Depending on the value of the exponent value exp, the time counter digit is shifted to perform the operation.

When the TAT, I, and L parameters are set from a firmware or the like, the relationship of the decimal point positions is adjusted in advance. The absolute value of the decimal point position is adjusted by the digit shift of the time counter.

In FIG. 8, the exponent value exp is read from the memory 80 and digit-shifted in accordance with the magnitude of the exponent value exp in order to align digits at the time 65 for each connection obtained in FIGS. Step S80). This digit shifted current time value ta
Is the same as TAT.

Next, the difference between the time value ta and TAT is calculated (step S81). In this calculation result, TAT is further compared. If TAT <ta, since TAT is in the past, the cell flow is in an acceptable state, so it is determined as conforming (A) (step) S82).

On the other hand, if TAT is large, (TAT-ta) is compared with L (step S8).
3),
If ta ≧ TAT-L, the difference from TAT is within the allowable value L, so
B) is determined (step S84). Further, if ta <TAT-L, the allowable value L is exceeded, so that it is determined as non-conforming (step S85).

Further, in the case of the adaptation (a), the current time ta is written as the next TAT, and in the case of the adaptation (b), the value increased by the parameter I and (TAT + I) are written back to the memory 62 as updated values.

In conventional floating point arithmetic, each time a parameter is read from the memory, the decimal point position must be adjusted, and when writing back to the memory, a procedure for decomposing it into an exponent part and a mantissa part is necessary. It was.

On the other hand, with the above configuration, in the present invention, the digit shift of the time counter is performed only once in the calculation. In addition, since the decimal point position can be changed with the exponent value exp for each connection, the calculation can be performed without reducing accuracy.

[Overflow processing in semi-fixed point arithmetic UPC processing]
Here, in the configuration of FIG. 7 described above, the decimal point positions of the TAT and the time counter are different for each connection. For this reason, the above-described method of subtracting a constant from TAT cannot perform overflow processing for each connection.

Therefore, as shown in FIG. 9, the time counter length (maximum value of the time counter + 1) is held as a constant (C) 67. The held value is shifted by the exponent value exp (step 90) to obtain a decrement value 68 of TAT69. The decrement value 68 is decremented from the TAT (step 91).

Although the shift process (step S90) in FIG. 9 and the shift process (step S81) in FIG. 8 have different purposes, the hardware can be shared because they have the same function.

In FIG. 9, the configuration is made without considering the multiplexing of a plurality of physical lines, but when applied to multiplexing, it can be easily achieved by combining FIG. 6 or FIG.

[Example]
Next, a configuration example of a UPC device to which the above-described features of the present invention are applied will be described. FIG. 11 is a block diagram of an embodiment to which the above-described features are applied as an embodiment of the present invention, and is based on the multiplexing of physical lines shown in FIG.

In FIG. 10, the three physical lines a, b, and c of the three physical layer termination devices I / F 1 to I / F 3 are multiplexed by the line multiplexer 21 and sent to the UPC device 20. Each physical line includes a plurality of connections as described above.

Further, in the line multiplexer 21, when cells from each line are multiplexed, the corresponding line identifiers a, b, c are added and sent to the UPC device 20 each time a cell is input. If the speed after multiplexing is high, the invalid cell X is multiplexed and inserted for speed adjustment.

Assuming such multiplexing of physical lines a, b, and c shown in FIG. 10, U in FIG.
The configuration of the PC device 21 will be described.

The line identifier 60 and the connection identifier 63 are grasped from the input cell CL. The time counter 200 is selected by the selector 201 based on the line identifier 60.
The time counter a for the corresponding physical line is selectively incremented by +1 and stepped.

The count value 61 of the time counter a after adding +1 indicates the time of the physical line.

If a cell of a specific physical line does not come, the time counter of that line will not advance.
Therefore, no problem occurs even if physical lines having different speeds are multiplexed. The UPC process and the counter overflow process are not performed simultaneously, but one is processed and the other is processed.

Now, description will be made assuming that the overflow processing is performed first. In the overflow process, the TAT memory 62 uses the counter value after addition as an index (address).
Read TAT and line number from.

That is, in the TAT memory 62, as described above, for each connection, that is, T
The number of the line to which the connection belongs is stored for each AT. Therefore, the TAT 620 and the line number 621 are read from the TAT memory 62 based on the counter value 61 after the addition.

At the same time, the exponent value exp corresponding to the increment information I to be accessed is read from the memory 100 by the counter value 61 after the addition.

Here, if the connection is not related to the stepped time counter, that is, the connection belongs to the corresponding line, the overflow process is not necessary. For this purpose, the comparator 102 compares the line number 621 stored in association with the TAT and the line identification number 6 of the input cell. In this comparison, if they do not match, the TAT update is inhibited (see FIGS. 11 and 103).

If the TAT update matches, then TAT is changed to TAT ′ as shown in FIG.
Update to For this purpose, the subtraction constant (Clen) 104 is shifted from the TAT by the shifter 105 according to the exponent value exp, and the value shifted by this shift is subtracted from the TAT 620 by the adder 106. The subtraction result 107 is updated to the memory 62 T
By writing back as AT ′, overflow processing is performed.

In the adder 106, a number obtained by adding 1 to the bit string obtained by inverting each bit of the output of the shifter 105 (2's complement to the output of the shifter 105).
A subtraction operation is performed by adding the value of TAT620 to.

Here, in FIG. 11, the shifter 105 and the adder 106 can be shared with the UPC process, as will be described later.

Further, the subtraction constant 104 to be shifted by the shifter 105 is determined in relation to the number of bits of the time counter 200. For example, when the time counter 200 has a length of 16 bits and the specified decimal point position is between bit 13 and bit 14 (exponential value exp = 0), a subtraction constant (16 bits + 13 bits = 29 bits) -229) is prepared.

If the exponent value exp of the connection to be overflowed is 5,
-2 (29-5) = 224 is obtained by the shifter 105. The overflow process is completed by subtracting this value from TAT 620 and writing back the result 107.

Next, the UPC process to be executed will be described. In the UPC process, UPC parameters (TAT, I, L) corresponding to the connections stored in the memories 62, 100, 101, which are referred to by the connection identifier 63, and decimal point position information, that is, the exponent value exp are used.

The timekeeping function necessary for the UPC process uses the current time ta 65 unique to the connection. The time ta 65 is obtained by subtracting the connection identifier (number) 3 by the adder 66 from the value 61 added by +1 when the cell is input.

Thus, even at the same time, the time ta 65 has a different value for each connection. At time ta 65, it is necessary to match the digits with TAT for UPC calculation. This digit alignment processing is realized by the shifter 105 shifting by the exponent value exp (121).

Since the shift result 107 may be the next TAT, the register 108
Temporarily held. A difference between the shifted time ta and TAT is obtained by an adder 106, and the magnitude relationship is determined by a sign.

In this determination process, if the minus sign, that is, TAT <ta, the input cell is determined to be compatible because the TAT is in the past. In this case, the cell is sent out to the network through the delay unit DL and the gate G.

At this time, the shifted ta held in the temporary memory 118 is changed to TAT.
As shown in FIG.

When TAT> ta, the cell has arrived earlier than the scheduled TAT.
The following judgment is necessary. That is, as a judgment, it is a problem how early the arrival is. For this determination, the comparator 109 compares the L value indicating the allowable range with the magnitude relationship of TAT-ta.

If L ≧ TAT−ta, it is determined that the value is within the allowable range, and TAT + I is written back to the memory 62 as the next TAT. This TAT + I is obtained by adding I read from the memory 100 to the TAT read from the memory 62 by the adder 112. Also at this time, the input cell is sent to the network through the delay unit DL and the gate G.

Further, if L <TAT-ta, the cell CL has arrived earlier than the allowable value L with respect to the scheduled arrival time TAT. in this case,
The TAT update process is not performed, and the gate G is closed and discarded for the cell CL.

Although the embodiments of the present invention have been described above with reference to the drawings, in the present invention, even if a plurality of physical lines are multiplexed, cell fluctuation due to multiplexing does not occur. Even if physical lines having different speeds are multiplexed, independent UPCs can be configured.

Even if a plurality of physical lines are multiplexed, there is no influence on UPC processing by other physical lines. In addition, the sum of the transmission speeds on the physical line and the U due to the difference in speed after multiplexing.
There is little PC error.

Furthermore, fixed point arithmetic can be applied to a wide range of transmission rates. For this reason,
It can be configured with high speed and small hardware.

It is a figure explaining the principle of the 1st feature point of the UPC apparatus of this invention. It is a time chart explaining the operation | movement of FIG. It is a figure explaining the principle of another characteristic of this invention. It is a figure explaining the principle of another characteristic of this invention. It is a figure explaining the time counter for every connection which belongs to the same physical line. FIG. 6 is a diagram (part 1) for explaining the feature of the present invention to which FIG. 5 is applied in a configuration using a counter for physical lines; FIG. 6 is a diagram (part 2) for explaining the feature of the present invention to which FIG. 5 is applied in a configuration using a counter for physical lines; As an example according to the present invention, a UPU process is performed in VS using a floating point arithmetic method. It is a figure explaining the overflow process in a semi-fixed point arithmetic UPC process according to this invention. It is a figure explaining the case where 3 physical lines are multiplexed as an Example to which the characteristic of this invention is applied. FIG. 11 is an embodiment to which the features of the present invention based on FIG. 10 are applied, and is a configuration example block diagram including a processing flow. It is a block diagram of a configuration example of an ATM switching system. It is a figure explaining the relationship between a multiplexing apparatus and a UPC apparatus at the time of assuming the case where two subscriber lines are multiplexed. It is a time chart corresponding to FIG. 13, and is a figure explaining the problem of the conventional line multiplexing. It is a figure explaining the problem that UPC precision is restrict | limited when the sum of the speed | rate after multiplexing and the speed of each interface circuit does not become an appropriate ratio. There is.

Explanation of symbols

1 ATM switch part 2 Input line correspondence part 3 Output line correspondence part 20 UPC device
21 Line multiplexer 22 Header converter 23, 31 OAM circuit
30 buffer circuit 10 ATM cell 11 header part 200 time counter 201 line multiplexing part 60 line number register 61 counter value after addition 62 TAT memory 63 connection identifier 65 time by connection

Claims (2)

In a cell flow rate control method in a cell switching system for receiving cells sent from a plurality of lines and controlling a cell path,
Executed by one common processing circuit ,
Multiplexing the plurality of lines;
A step of incrementing only the time counter corresponding to the line to which the incoming cell belongs among the plurality of time counters provided for the plurality of lines to be multiplexed ;
And a step of controlling the cell flow rate corresponding to the count value of the stepped time counter,
As a step of controlling the cell flow rate,
Storing a parameter as a reference for controlling the cell flow rate for connections belonging to each of the plurality of lines in a table;
The parameters are stored in fixed point, and decimal point position of the parameters are recorded in a typical one index value for each connection,
When calculating the parameter and the count value of the time counter, the count value of the time counter is shifted by the exponent value and calculated.
A cell flow rate control method. In a cell switching system that receives cells sent from multiple lines and controls the cell path,
Configured as one processing circuit,
A line multiplexing unit for multiplexing the plurality of lines;
A plurality of time counters provided only for a plurality of lines multiplexed by the line multiplexing unit and incremented only by a time counter corresponding to a line to which an incoming cell belongs ;
And a cell flow rate control device for controlling the cell flow rate according to the count value of the stepped time counter ,
The cell flow controller is
A table for storing parameters serving as a reference for controlling the cell flow rate for connections belonging to each of the plurality of lines;
The parameters are stored in fixed point, and decimal point position of the parameters are recorded in a typical one index value for each connection,
When calculating the parameter and the count value of the time counter, the count value of the time counter is shifted by the exponent value and calculated.
A cell switching system characterized by that.

Images