JP3602893B2 - ATM interface and shaping method - Google Patents

Description

[0001]
[Industrial applications]
The present invention relates to an interface device and a traffic control method for an asynchronous transfer mode (ATM) network, and more particularly to a shaping technique for controlling transmission of cells according to declared traffic.
[0002]
[Prior art]
In the network of the ATM switching system, by using fixed-length packets called cells, network resources are shared by a plurality of connections, thereby enabling high-speed and efficient transmission. The cell is composed of, for example, a 53-byte length including a 5-byte header section and a 48-byte information section. The header section includes an identifier VPI of a virtual path (hereinafter, referred to as a VP) and multiplexed with the VP. And an identifier VCI of a virtual channel (hereinafter, referred to as VC) to be executed.
[0003]
Regarding the traffic control of the ATM network, for example, ITU-T, Draft Recommendation I.T. At 373, at the time of call origination, the user (source device) declares traffic parameters such as communication speed and communication quality for the traffic to be set up, and allocates resources for each connection based on the report parameters. "Control" and "Usage parameter control (hereinafter referred to as" control ") to monitor the status of input cells in order to guarantee communication quality and take measures such as marking and cell discarding for cells transmitted in violation of declared traffic. , UPC) ").
[0004]
An ATM interface provided in a user terminal or a user network interface device (hereinafter referred to as UNI) has a traffic control function to output cells to the ATM network so as not to violate the declared traffic in order to avoid the above-mentioned cell discard by UPC. "Shaping function" is required.
The above literature describes a leaky bucket algorithm and a virtual scheduling algorithm as shaping control algorithms applicable to UPC.
As another prior art related to traffic control, for example, Japanese Patent Application Laid-Open No. 5-130136 proposes a "transmission bit rate monitoring method" in which shaping with UPC is performed by the apparatus configuration shown in FIG.
[0005]
Here, focusing on only the shaping function, in the configuration of FIG. 12, when a cell transmitted from the user apparatus and passing through, for example, an ATM multiplexer arrives at the apparatus 500, the header identification unit 501 detects the input cell. And notifies the time calculation unit 502 of the VCI, and transfers the input cell to the memory device 503. The time calculation unit 502 calculates a waiting time D of a cell in the memory device using a leaky bucket algorithm so as to protect traffic declared in advance for each VCI. The input cells stored in the memory device 503 are read from the memory device in response to the instruction from the time calculation unit 502 when the respective waiting times D have elapsed, and sent out to the output line r.
[0006]
[Problems to be solved by the invention]
However, since each exchange configuring the ATM network manages the bandwidth in units of VP, when each user uses the public ATM network, not only the traffic corresponding to the VCI but also the traffic corresponding to the VPI. You need to declare.
However, in the conventional shaping control, the cell output control is performed according to the reporting traffic for each VCI. For example, when the reporting peak rate corresponding to the VPI is lower than the output interface speed, the reporting traffic for the VCI is protected. However, when observed for each VPI, the cell transmission interval may be faster than the declared peak rate, resulting in a violation of the declared traffic.
[0007]
FIG. 13 shows an example of a cell transmission interval by the conventional shaping control.
In the figure, "TA" is a reporting peak interval in a virtual path having VPI = "A", and T (a) and T (b) are VCI = "a" and VCI = "" formed on the virtual path, respectively. b shows the declared peak interval of the virtual channel with "b". Reference numerals 601 and 602 denote cells belonging to the virtual channel of VCI = “a”, and 611 and 612 denote cells belonging to the virtual channel of VCI = “b”.
In the illustrated example, the transmission interval between the cells 601 and 602 and the transmission interval between the cells 611 and 612 satisfy the respective declared values T (a) and T (b). However, when observing the interval between cells having VPI = “A” in units of virtual paths, the transmission interval between cells 602 and 612 violates the declared peak interval TA.
[0008]
An object of the present invention is to provide an improved shaping method and an ATM interface in an ATM network in which a plurality of virtual channels are formed on one virtual path.
It is another object of the present invention to provide a shaping method and an ATM interface capable of satisfying declared traffic in both VPI and VCI.
Still another object of the present invention is to provide an ATM interface device applied to a network in which a plurality of VPs are multiplexed on one physical line, or to make the declared peak rate corresponding to the VPI slower than the output line speed. An object of the present invention is to provide a shaping method and a control device capable of controlling a cell transmission interval in accordance with a reporting traffic for each VPI / VCI and a reporting peak rate of a VPI in an ATM interface device which handles various input cells.
[0009]
[Means for Solving the Problems]
In order to achieve the above object, a shaping method for controlling a transmission interval of a cell to a transmission line according to the present invention includes the steps of temporarily storing input cells in a buffer memory and identifying an identifier of a group to which the input cells belong. The transmission time of the input cell is determined in accordance with both the traffic condition previously declared in correspondence with the traffic condition and the traffic condition previously declared in correspondence with the identifier of the subgroup to which the input cell belongs. And comparing the transmission time of the cell with the transmission time assigned to the first-arrived cell. If the transmission times overlap, the transmission time determined in the first step is corrected, and then the transmission time is corrected. And a second step of storing a correspondence between the cell identifier and the input cell identification information, and a correspondence between the cell identifier and the transmission time stored in the second step. It reads the cells stored in the serial buffer memory to the delivery order of time, characterized in that and a third step of transmitting to the output line.
[0010]
Further, the ATM interface according to the present invention includes a buffer memory for temporarily storing a plurality of ATM cells input from an input line, writing of cells to the buffer memory, and writing from the buffer memory to the output line. Control means for reading the cell, the control means,
First table means for storing control parameters obtained corresponding to traffic conditions previously declared for each group and subgroup to which an input cell belongs, and for storing empty states corresponding to time slots on the output line And a second table means for transmitting the cell when the cell arrives from the input line, based on the control parameters corresponding to the group and subgroup to which the cell belongs stored in the first table means. The timing is obtained, an empty transmission time slot to be associated with the transmission timing is determined with reference to the second table means, and the corresponding cell stored in the buffer memory is determined at the timing of the transmission time slot. Access means for reading out to the output line.
More specifically, the above group corresponds to, for example, a virtual path (VP) formed by multiplexing on a transmission path, and the subgroup corresponds to a virtual path formed by multiplexing on each virtual path. The channel (VC) corresponds. The control parameters stored in the first table means include, for example, a peak cell interval corresponding to a declared peak rate for each group (virtual path identifier: VPI), and a subcell (virtual channel identifier: VCI) for each group. Is the peak cell interval corresponding to the declared traffic.
[0011]
In the preferred embodiment of the present invention, the cell arrival time and the cell transmission time slot are managed using the time required to transfer one cell at the output interface speed as one unit. In this case, the first table means stores the peak cell interval for each group (virtual path identifier: VPI) as a value in which one cell transfer time at the output interface speed is one unit. The peak cell interval for each group (virtual channel identifier: VCI) is stored as a value with the peak cell interval for the group (VPI) as one unit. At the time of cell arrival, the relative cell transmission time is calculated based on the VPI peak cell interval as one unit according to the control parameter for each subgroup (VCI), and the idle time is calculated by referring to the second table means. A cell transmission time slot to be associated with the relative cell transmission time is selected from the slots.
[0012]
According to the above configuration, since the use state (empty state) of the time slot by another cell is stored in the second table means, the relative cell transmission time is temporarily changed to the cell belonging to another subgroup in the same group. Alternatively, when there is a conflict with the relative cell transmission time (time slot) of a cell belonging to another group, the time can be shifted to a new relative cell transmission time that does not compete with other cells.
The second table is, for example, a first bitmap composed of a plurality of bit positions corresponding to times (time slots) used for storing the empty state of the relative cell transmission time for each group, and is common to all groups. And a second bit map including a plurality of bit positions used for storing the time slot empty state.
[0013]
[Action]
According to the present invention, it is possible to determine a cell transmission time (time slot) according to a report parameter for each VCI while keeping the peak cell interval for each VPI, and the transmission time obtained from the control parameter is used as the transmission time for another cell. Even when there is a conflict, the transmission timing can be determined within the free time (time slot) within a range that does not violate the declared peak cell interval for each VPI, so that the cell is read from the buffer memory at this transmission timing. By doing so, shaping for each VPI and for each VCI can be realized.
[0014]
【Example】
FIGS. 2A and 2B show an example of a configuration of a communication system to which the ATM interfaces 1a to 1c according to the present invention are applied.
FIG. 2A shows an example in which the ATM interfaces 1a and 1b are applied between the private ATM switch 2 and the wide area ATM network.
The ATM interface 1a is connected between one output line 4a of the ATM exchange 2 and one input line (subscriber line) 5a of the wide area ATM network, and the output line 4a is connected to the interface speed r of the wide area ATM network input line 5a. The ATM interface 1a relays the cells belonging to the same VP output from the ATM switch 2 to the output line 4a to the input line 5a.
[0015]
The ATM interface 1b is connected between one output line 4b of the ATM switch 2 and a plurality of input lines 5b-1 to 5b-N of the wide area ATM network. The ATM interface 1b is multiplexed and output from the ATM switch 2 to the output line 4b. And receives the cells having different VPs and distributes them to the input lines 5b-1 to 5b-N corresponding to the VPs.
FIG. 2B shows an example in which the ATM interface 1c is connected between the ATM multiplexer 3 in the user premises and the wide area ATM network. The ATM interface 1c receives a plurality of cells having different VPs multiplexed from the multiplexer 3 and output to the output line 4c, and relays the cells to the input line 5c to the wide area ATM network. In this case, the output line 4c and the input line 5c of the wide area ATM network have the same interface speed.
[0016]
FIG. 3 shows functional blocks of the ATM interface 1a for shaping cells of the same VP.
FIG. 3A shows that the input cell received at the interface speed r from the line 4a is converted into the output-side interface speed r 'by the speed conversion means 11a using a buffer memory such as a FIFO, and then a report is made for each VPI. A configuration is shown in which shaping is performed by a shaping means 10a having a VP shaping function according to traffic and a VC shaping function according to declared traffic for each VCI, and output to a line 5a.
In FIG. 3B, a buffer for speed conversion and a buffer for VP shaping and VC shaping are shared, and cells received from the line 4a are processed by a shaping circuit 10b having a speed conversion function. The configuration is shown.
[0017]
FIG. 4 shows a functional block of an ATM interface 1b for separating an input cell into a plurality of lines corresponding to VP.
In FIG. 4A, the input cells of the interface speed r received from the line 4b are sorted by VPI by the separation circuit (selector) 12a, and the speed r 'is set by the speed conversion means 11b-1 to 11b-N provided corresponding to the VPI. After the conversion into the cell stream of 1 to r'N, the shaping circuits 10c-1 to 10c-N simultaneously perform VC shaping and VP shaping, and output the shaped cells to the lines 5b-1 to 5b-N. The following is the configuration.
FIG. 4B shows that the cells sorted by VPI in the separation circuit 12b are input to shaping circuits 10d-1 to 10d-N having a speed conversion function provided for the VPI, and the VPs are simultaneously converted. A configuration is shown in which shaping and VC shaping are performed and then output to VPI-compatible lines 5b-1 to 5b-N.
FIG. 4C shows a configuration in which an input cell from the line 4b is subjected to VP shaping and VC shaping by the shaping circuit 10e, and then distributed to the VPI-compatible lines 5b-1 to 5b-N by the separation circuit 12c. Show.
[0018]
FIG. 5 shows a functional block diagram of an ATM interface 1c that multiplexes input cells of a plurality of VPs multiplexed and input from the line 4c to the line 5c and outputs the multiplexed cells. In this case, the cells received from the line 4c are subjected to VP shaping and VC shaping by the shaping circuit 10f, and then output to the line 5c at the same interface speed r as the input line 4c.
[0019]
Next, the configuration of the shaping circuit 10 will be described. Here, the configuration of the shaping circuits 10e and 10f that handle input cells of different plural VPs will be described. However, the shaping circuits 10b and 10d that handle input cells of the same VP and the shaping circuits 10a and 10c separately provided with a buffer for speed conversion are provided. Is easily obtained from the circuit configuration described below, and the description is omitted.
[0020]
FIG. 1 is a block diagram showing an example of the configuration of the shaping means 10.
The shaping circuit 10 includes a cell buffer 20 for temporarily storing input cells, a header identification unit 30 for identifying VPI / VCI included in a header of the input cell, and a transmission time (transmission time slot) of the input cell. ), A vacant time search unit 50 for searching for a vacant time (vacant time slot) corresponding to the transmission time calculated by the above-mentioned transmission time calculation unit 40, and a cell buffer 20. A buffer control unit 60 for controlling writing and reading of cells to and from a parameter table 70 for storing various parameters such as a report traffic and a cell transmission time for each connection, and a transmission time state (vacancy / occlusion). A search table 80 for storing state information indicating the state of the cell to be read from the cell buffer 60. Composed of cell output list 90. for storing the file number (buffer address).
[0021]
Cells input from the line 4 at the interface speed r are sequentially written into the cell buffer 20 and temporarily stored until the transmission time comes. At this time, the VPI / VCI included in the header of the received cell is identified by the header identification unit 30, and the identified VPI / VCI is notified to the transmission time calculation unit 40 via the bus 5.
The transmission time calculation unit 40 calculates the input transmission time by using the parameters stored in the parameter table 70 in advance so that the cell traffic flows in the declared traffic for each VPI and VCI, and calculates the calculation result (transmission result). (Time) to the free time search unit 50 via the bus 6.
The vacant time search unit 50 refers to the search table 70 to search for a vacant time (time slot) in which cells can be transmitted in the transmission time notified by the transmission time calculation unit 40 or in a time zone thereafter. The detected free time is notified to the buffer control unit 60 via the bus 7.
[0022]
The buffer control unit 60 stores, in an entry corresponding to the vacant time (time slot) in the cell output list 90, identification information of a cell to be transmitted at the vacant time (for example, a buffer address or a cell indicating one accumulation of the cell). Buffer number). The buffer control unit 60 reads out the cell identification information corresponding to the time slot from the cell output list 90 at each time slot on the output line 5 having the interface speed r ', and based on the cell identification information, The cell is read from 20 and output to line 5.
[0023]
FIG. 6 shows an example of the contents of the parameter table 70 referred to by the transmission time calculation unit 40.
The parameter table 70 includes a first parameter table 71 for storing a peak cell interval T expressed on the basis of the interface speed r ′ for VPI, and a peak rate of each VPI for VPI / VCI. And a second parameter table 72 for storing an ideal transmission time tn '.
Here, the arrival time of the cell is expressed as one unit of the transfer time of one cell at the interface speed r ′, and as an example, the declared traffic (peak rate) of VPI = “1”, “A”, “N” is respectively "R'1", "r'A", "r'N", and the declared traffic (peak rate) of VPI / VCI = (1), (a), (N) is "P (1)", respectively. , “P (a)” and “P (m)” are shown. When all the input cells supplied to the shaping circuit 10 have the same VPI, a register that stores one peak cell interval can be applied instead of the first parameter table 71.
[0024]
FIG. 7 is a flowchart showing a procedure for calculating the transmission time performed with reference to the parameter table 70.
Here, assuming that the cell arrival time is the current time ta indicated by the timer, the VPI value notified from the header identification unit 30 is “A”, and the VPI / VCI value is “(a)”, the transmission time calculation unit When the VPI and VCI are received from the header identification unit 30 via the bus 5 (step 101), the 40 accesses the first parameter table 71 using the VPI value “A” as an address, and obtains the peak cell interval “ TA ”is read and the arrival time ta is divided by TA to convert the arrival time ta to a value in units of the cell transmission timing at the peak rate, and this is set as the control arrival time“ ta′A ”( Step 102).
[0025]
Next, the second parameter table 72 is accessed using the VPI / VCI (= (a)) as an address, and the ideal transmission time “tn ′ (a)” corresponding to the VPI / VCI is read, and the “ta′A” is read. (Step 103). If the ideal transmission time “tn ′ (a)” is earlier than the arrival time “ta′A” (tn ′ (a) <ta′A), the ideal transmission time tn ′ (a) = ta′A, and ta ′ A value obtained by rounding up the decimal part of A is set as the transmission time to'A (step 104). If tn ′ (a) ≧ ta′A, the value obtained by rounding up the decimal part of the ideal transmission time tn ′ (a) is set as the transmission time to′A (step 105). When the value of the transmission time to'A is determined, the transmission time to'A is notified to the idle time search unit 50 via the bus 6 (step 106), and the value of the ideal transmission time stored in the second parameter table 72 is set. Is updated to tn ′ (a) = tn ′ (a) + T ′ (a) (107).
[0026]
FIG. 8 shows an example of the shaping operation according to the present invention based on the above-described transmission time calculation.
Here, the peak cell interval TA of VPI (= A) is “3”, and the value of the peak cell interval of VPI / VCI = (a) and VPI / VCI = (b) belonging to this VPI is T ′ (a) = 4. , T ′ (b) = 3, the initial value 131 of the ideal transmission time tn (a) is “0”, and the initial value 231 of the tn (b) is “0”.
[0027]
Assuming that the arrival times ta of the cells 111, 112 and 113 of VPI / VCI = (a) are “0”, “17” and “24”, respectively, the arrival time ta′A for control of the first input cell 111 is Since the value is “0” as indicated by 121, the value of the transmission time to′A is “0” as indicated by 301, and the value of the ideal transmission time tn ′ (a) of the next cell is 132 Is updated to "4" as shown in FIG.
In the next input cell 112, the value of the arrival time ta′A for control is “17/3” as indicated by 122. Since this is greater than the ideal transmission time tn ′ (a) = “4”, the value of the transmission time to′A becomes “6” as indicated by 302, and the value of the ideal transmission time tn ′ (a) of the next cell Is updated to "29/3" as shown in 133.
In addition, in the cell 113, the value of the control arrival time ta'A becomes "8" as shown by 123, which means that the cell has arrived earlier than the ideal transmission time shown by 133. Therefore, the value of the transmission time to′A is “10” as indicated by 303, and the value of the ideal transmission time tn ′ (a) of the next cell is “4/3” as indicated by 134.
[0028]
On the other hand, assuming that the arrival times ta of the cells 211 and 212 of VPI / VCI = (b) are “1” and “9”, respectively, the value of the arrival time ta′A for control in the cell 211 is as indicated by 221. Therefore, the value of the transmission time to'A is "1" as shown in 311 and the value of the ideal transmission time tn '(b) of the next cell is "1/3" as shown in 232. 10/3 ". Further, in the next cell 212, the value of the control arrival time ta′A is “3” as shown at 222 and is earlier than the ideal transmission time value “10/3” shown at 232, The value of the time to'A is “4” as indicated by 312.
According to the present embodiment, not only the peak cell interval for each connection but also the VPI-compatible peak cell interval satisfies the declared connection condition from the transmission time value to'A described above.
[0029]
FIG. 9 shows an example of the configuration of the free time search unit 50 and the search table 80.
Reference numeral 81 denotes a VPI-specific search table composed of a plurality of VPI-compatible table areas 81-1 to 81-N. Each table area is composed of a plurality of bit areas corresponding to a cell transmission time (transmission time slot). For the VPI, a flag bit indicating whether each transmission time is empty or closed is stored.
[0030]
Reference numeral 82 denotes an all-connection search table including a plurality of bit areas corresponding to cell transmission times, and stores, as a flag bit, a vacancy / occupation state at each transmission time when viewed as a whole connection. In these search tables, for example, if there is a cell to be transmitted in the time slot of transmission time = i, “1” is set to the i-th bit of the table.
[0031]
When the free time search unit 50 receives the value of the transmission time to'A and the value of the peak cell interval TA from the transmission time calculation unit 40 via the buses 6a and 6b, the first search unit 51 stores the values in the VPI-specific search table 81. The table area 81-A is accessed, and from the time slots located after the transmission time to'A, a vacant time tgo'A whose flag bit is in the "0" state is searched for and passed to the calculation unit 52. . The calculation unit 52 calculates the transmission time to (= tgo'A × TA) based on the transmission time slot from the time tgo'A, and passes this to the second search unit 53. The second search unit 53 accesses the all connection search table 82 based on the transmission time to, searches for an idle transmission time tgo in a time slot after the time to, and transmits this via the bus 7. To the buffer controller 60.
[0032]
In the search for the flag bits indicating the empty state in the table areas 81 and 82, if a method of checking one bit at a time from the top of the table is used, the search takes a long time when there are many closed bits, so that several bits are required. And collectively retrieve the flag information, and perform a batch search using a priority encoder or the like. In order to further reduce the number of times of memory access, for example, a block-by-block state display register that indicates the occupied state of the table area for several bits by 1 bit may be provided. For example, when j bits of the memory 82 are made into one block and all time slots from the transmission time i to i + j are closed, “1” is set to the i / j-th register 54.
[0033]
FIG. 10 shows an example of an empty time search operation to which the block-by-block status display register 54 is applied.
When the value of the transmission time is to'A = “6” and the value of the peak cell interval corresponding to the VPI is TA = “4”, the sixth and subsequent bits of the memory 81-A are searched and “0” is set. The first bit position i is set as the actual transmission time tgo'A (arrow 401). At this time, the i-th bit of the memory 81-A is changed to “1”. Assuming that the value of tgo'A thus found is, for example, i = “7”, the calculation unit 52 multiplies the value “7” by TA = “4” and calculates to = “28”. calculate.
Next, a search is made for the 28th bit and subsequent bits of the memory 82 (arrow 402). If the register 54 stores the state of each block with 10 bits of the memory 82 as one block, if no bit of the “0” state is found up to 29 bits of the memory 82, the register 54 A search is made for 30/10 = 3rd and subsequent 54 of 54 (arrow 403). In the illustrated example, since the fifth bit of the register 54 is in the “0” state, the memory 82 is searched for the (5-1) × 10 = 40th bit and subsequent bits (arrow 404), and the first empty bit is searched. The position i (“41” in this example) is set as the value of tgo, and the flag “1” indicating the closed state is set at the ith bit of the memory 82.
[0034]
If it is desired to perform strict shaping also on a CDV (Cell Delay Variation) generated by a free time search, the value obtained by dividing the actual transmission time tgo by the peak cell interval TA of the VPI is the ideal transmission time in the second parameter table. The flag may be updated to “1” at a bit position corresponding to a value obtained by rounding up the decimal part of the above value in the table area 81-A in the first search table 81 by setting tn1 ′ (a).
[0035]
FIG. 11 shows an example of the configuration of the cell buffer 20, the buffer control unit 60, and the cell output list 90.
The cell buffer 20 is composed of a plurality of buffer areas in units of cells, in which write operations are performed sequentially and read operations are performed randomly. The buffer control unit 60 includes a write control unit 61 that generates a write address of the cell buffer 20, a transmission time registration unit 62 that registers a cell buffer number in the cell output list 90, and a read that generates a read address of the cell buffer 20. And a control unit 63. The cell output list memory 90 stores buffer numbers corresponding to transmission time slots.
[0036]
When an input cell is received from the line 4, the write controller 61 generates a sequential write address according to a buffer number output from a counter (not shown), and writes the input cell into the cell buffer 20. The transmission time registration unit 62 registers the write buffer number in an entry area of the cell output list 90 corresponding to the time tgo notified from the empty cell search 50 via the bus 7. For example, when tgo = “41” and the write buffer number = “17”, the buffer number “17” is written in the 41st entry of the cell output list 90.
The read control unit 63 reads the buffer number from the cell output list 90 according to the transmission time slot (for example, when the transmission time slot = “24”, the buffer number “4” is read), and reads according to the buffer number. An address is generated, one cell is read from the cell buffer 20 and transmitted to the line 5.
[0037]
In the above embodiment, a search table is prepared for each VPI, which indicates the occupied state at each time in the unit of the peak cell transmission interval by one bit, and the vacant time is searched when the transmission time conflicts. Strict shaping can be performed in CDV generation at the time of contention.
[0038]
FIG. 14 shows a specific configuration of the shaping circuit shown in FIG.
The timer 41 indicates the current time ta in units of one cell transfer time at the interface speed r of the input line 4. The transmission time slot generation 63a generates a transmission time slot number in units of one cell transfer time at the interface speed r 'of the output line 5. The selector 62 switches between a write address and a read address according to the R / W signal.
[0039]
First, the operation at the time of cell reception will be described. When an input cell is received from the line 4, the counter 61a counts up, and the WA generation circuit 61b generates a write address using the value of the counter 61a as a write buffer number of the buffer 20, and writes the received cell into the buffer 20.
At this time, the VPI identification circuit 31 and the VPI / VCI identification circuit 32 identify the VPI value and the VPI / VCI value from the header of the input cell. The peak cell interval T is read from the first parameter table 71 based on the VPI value output from the VPI identification 31, and the dividing circuit 42 divides the cell arrival time ta indicated by the timer 41a by the peak cell interval T for control. Arrival time ta 'is calculated. Further, based on the VPI / VCI value output from the VPI / VCI identification 32, the ideal transmission time tn 'is read from the second parameter table 72, and the comparator 43 determines the ideal transmission time tn' and the arrival time. ta ′ are compared. The selector 44 selects ta ′ if tn ′ <ta ′, selects tn ′ if tn ′ ≧ ta ′, and selects tn ′ according to the output of the comparator. I do.
[0040]
Based on the VPI output from the VPI identification circuit 31, the selector 51a selects a table corresponding to the VPI value from the VPI-based search tables 81-1 to 81-N. The contents of the above table are input to the priority encoder 51b, and the empty time tgo 'in which the flag bit is in the "0" state is collectively searched from the time slots located after the cell transmission time to', By multiplying by the peak cell interval T corresponding to the identification VPI value by the multiplication circuit 52, the transmission time to is calculated based on the transmission time slot.
[0041]
The transmission time to is divided by the number of bits (the number of slots) j for one block of the entire connection search table memory 82 by the division circuit 53a. Using the value of the integer part of to / j output from the division circuit 53a as a read address, one block of data is read from the entire VPI / VCI memory 82 and input to the priority encoder 53d. Also, a value indicating the remainder of the division circuit 53a is input to the priority encoder 53d, and the priority encoder 53d sets a flag bit after the bit indicated by the remainder value from the data of one block read from the memory 82. Is collectively searched for an empty bit position where is "0". If there is no empty bit in the data of one block, the enable signal is held in the latch 53e, and the selector 53b and the selector 53c switch their selection inputs. As a result, from the block status register 54, the bit position where the flag bit is in the “0” state after the bit position indicated by the integer part of to / j is collectively searched by the priority encoder 53f, and the found empty bit position is read out. The next one block of data is read from the table memory 82 as an address, and the priority encoder 53d searches for an empty bit position again.
[0042]
When a free bit position is found by the priority encoder 53d in this way, a value indicating the free bit position is input to the adder 53h, and is added to a value obtained by multiplying the read address of the memory 82 calculated by the multiplier 53g by j. Is output as a value indicating the actual transmission time tgo. The write buffer number indicated by the counter 61a is registered in the cell output list 90 using the value of the actual transmission time tgo as a write address.
[0043]
The empty bit position found by the priority encoder 53d is decoded by the decoder 55a, and the flag bit "1" is set in the above-mentioned bit position of the all connection search table 82 by the flag update circuit 55b. Further, the comparator 55c checks whether or not all the flag bits of one block including the bit position have become “1”. If all the flag bits are “1”, the read address of the table 82 is decoded by the decoder 55d. Then, the flag at the bit position to which the block belongs in the register 54 is changed to "1" by the flag update circuit 55e.
[0044]
The actual transmission time tgo output from the adder 53h is also input to the divider 56a, and the ideal transmission time tn 'is calculated by dividing tgo by the peak cell interval T of the VPI. The calculated ideal transmission time tn 'is input to the decoder 56b, and a bit position corresponding to a value obtained by rounding up a decimal part from the value of the ideal transmission time tn' is input to the flag updating circuit 56c. The flag of the table corresponding to the VPI value of the input cell is updated. The value of tn 'output from the divider 56a is also supplied to the adder circuit 45, and the peak cell interval T' of VPI / VCI is added to the value of tn 'to obtain the ideal value of the second parameter table 72. The value of the sending time is updated.
[0045]
The cell transmission operation is performed as follows. In each transmission time slot, a transmission time slot number is output from the transmission time slot generation circuit 63a. The transmission time slot number is provided to the cell output list 90 via the selector 62 as a read address. As a result, the buffer number registered in the time slot is read from the cell output list 90 and supplied to the read address (RA) generation circuit 63b. The RA generation circuit 63b generates a read address based on the buffer number, reads a cell from the buffer 20, and sends the cell to the line 5.
[0046]
【The invention's effect】
As is apparent from the above description, according to the present invention, in consideration of both VPI and VCI declaration parameters, for example, the transmission time is calculated so as to satisfy the VCI-compliant declaration parameters in units of VPI peak cell transmission intervals. By calculating the transmission time that satisfies the reporting peak rate of the VPI, it is possible to realize the shaping control that satisfies both the reporting traffic for each VCI and the reporting traffic for each VPI.
[Brief description of the drawings]
FIG. 1 is a block diagram showing one embodiment of an ATM interface according to the present invention.
FIG. 2 is a system configuration diagram showing an application example of an ATM interface according to the present invention.
FIG. 3 is a block diagram showing a functional configuration of an ATM interface for processing cells belonging to the same VP.
FIG. 4 is a block diagram showing a functional configuration of an ATM interface for processing cells of a plurality of VPs and relaying the cells to a plurality of output lines.
FIG. 5 is a block diagram showing a functional configuration of an ATM interface for processing cells belonging to a plurality of VPs and relaying the processed cells to one line.
FIG. 6 is a diagram showing a configuration of a parameter table in FIG. 1;
FIG. 7 is a flowchart showing an example of a transmission time calculation in FIG. 1;
FIG. 8 is a diagram showing an example of a shaping operation.
FIG. 9 is a diagram showing an example of a configuration of a free time search unit and a search table in FIG. 1;
FIG. 10 is a diagram showing an operation example of a free time search in FIG. 9;
FIG. 11 is a diagram showing an example of a configuration of a cell buffer, a buffer control unit, and a cell output list in FIG. 1;
FIG. 12 is a diagram showing a conventional example of a shaping function.
FIG. 13 is a diagram for explaining a problem in conventional shaping.
FIG. 14 is a diagram showing an example of a detailed configuration of a shaping circuit of FIG. 1;
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... ATM interface, 10 ... shaping circuit, 20 ... cell buffer, 40 ... transmission time calculation part, 50 ... empty time search part, 60 ... buffer control part,
70: parameter table, 80: search table, 90: cell output list.

Claims (18)

A shaping method for controlling a transmission interval of cells to a transmission path,
The input cells are temporarily stored in a buffer memory, and the traffic conditions previously declared corresponding to the identifier of the group to which the input cells belong, and the traffic conditions previously declared corresponding to the identifiers of the subgroups to which the input cells belong. A first step of determining the transmission time of the input cell in accordance with both of the traffic conditions being performed;
The transmission time of the cell is compared with the transmission time assigned to the first-arrived cell. If the transmission times overlap, the transmission time determined in the first step is corrected, and then the transmission time and the input cell are identified. Reading the cells stored in the buffer memory in the order of transmission time based on the correspondence between the cell identifier and the transmission time stored in the second step for storing the correspondence with the information; And a third step of transmitting to a line. The shaping method according to claim 1, wherein the traffic conditions corresponding to the group identifier and the subgroup identifier are respectively stored as peak cell intervals in an output line. 3. The shaping method according to claim 1, wherein the transmission time indicates a time slot defined in time series with one cell transfer time in an output line as a unit. 4. The method according to claim 1, wherein the identifier of the group is an identifier of a virtual path, and the identifier of the subgroup is an identifier of a virtual channel multiplexed on the virtual path. The shaping method described. A shaping method for controlling a transmission interval of cells to a transmission path in correspondence with a group and a subgroup to which each cell belongs,
The first peak cell interval corresponding to the declared peak speed is set for each group with one cell transfer time determined by the output interface speed as one unit, and the declared traffic for each subgroup is set with the first peak cell interval as one unit. Memorize the parameters corresponding to the speed,
When a cell arrives, a relative cell transmission time is determined based on a parameter corresponding to a subgroup to which each cell belongs, using the first peak cell interval corresponding to the cell as one unit, and is determined by the output interface speed from the relative cell transmission time. Determine a cell transmission time with one cell transfer time as one unit,
A shaping method, wherein each cell is transmitted according to the cell transmission time. The parameter includes a second peak cell interval corresponding to the declared peak speed for the subgroup;
At the time of cell arrival, a cell arrival time is obtained using the first peak cell interval of the group to which each cell belongs as one unit,
The relative cell transmission time is determined based on the cell arrival time, the transmission time of the immediately preceding cell in the subgroup to which the cell belongs, and the second peak cell interval. Shaping method. When the relative cell transmission time conflicts with the relative cell transmission time of a cell belonging to another subgroup in the same group, determining an empty relative cell transmission time existing after the relative cell transmission time of the cell. The shaping method according to claim 5 or 6, wherein: When the relative cell transmission time conflicts with the relative cell transmission time of a cell belonging to another subgroup in the same group, a new cell that does not compete with another cell in the group existing after the relative cell transmission time of the cell. The relative cell transmission time,
When the cell transmission time obtained for the new relative cell transmission time conflicts with the transmission time of a cell belonging to another group, another empty cell transmission time existing after the cell transmission time is obtained. The shaping method according to claim 5 or 6, wherein the shaping method is performed. The empty state of the relative cell transmission time is stored for each group by a first bitmap consisting of a plurality of bit positions corresponding to the time, and the cell transmission time is stored in a second bitmap consisting of a plurality of bit positions corresponding to the time. When the relative cell transmission time conflicts, the empty state existing after the bit position corresponding to the relative cell transmission time is referred to by referring to the first bitmap corresponding to the group to which the cell belongs. A state bit is searched, and a time corresponding to the bit is obtained as a relative cell transmission time.
When the cell transmission time conflicts, an empty bit existing after the bit position corresponding to the cell transmission time is searched with reference to the second bitmap, and the time corresponding to the bit is determined as the cell transmission time. The shaping method according to any one of claims 5 to 8, wherein: A register corresponding to the first bitmap or the second bitmap, wherein a plurality of bits in each bitmap are regarded as one block, and the state of each block is stored in the register;
10. The shaping method according to claim 9, wherein a reference range in the first bitmap or the second bitmap is specified with reference to the register. An empty time is searched for in a predetermined block of the first bitmap or the second bitmap, and if no empty time exists in the block, the first bitmap or the second bitmap is referred to by referring to the register. 11. The shaping method according to claim 10, wherein a next reference range in the map is specified. 12. The shaping method according to claim 5, wherein the group is a virtual path, and the subgroup is a virtual channel multiplexed on the virtual path. An ATM interface having a shaping function for controlling transmission of ATM cells to an output line in accordance with a group and a subgroup to which each cell belongs, and temporarily stores a plurality of ATM cells input from an input line. Buffer memory for
Control means for writing cells to the buffer memory and reading cells from the buffer memory to the output line, wherein the control means
With one cell transfer time determined by the cell speed on the output line as one unit, a first peak cell interval obtained corresponding to the declared peak speed for each group, and a subunit with the first peak cell interval as one unit Table means for storing control parameters obtained in accordance with the declared traffic speed for each group;
When a cell arrives from the input line, based on the control parameter corresponding to the subgroup to which the cell belongs stored in the table means, the relative cell transmission time is set with the first peak cell interval corresponding to the cell as one unit. Then, a cell transmission time is determined from the relative cell transmission time with one cell transfer time determined by the interface speed of the output line as one unit, and each cell stored in the buffer memory is determined according to the cell transmission time. And an access means for reading the data from the ATM interface. 14. The ATM interface according to claim 13, wherein the relative cell transmission time and the cell transmission time specify one of time slots determined by a band of the output line. The control means has a memory means for storing information specifying a cell to be read from the buffer memory in association with a time slot on the output line, and the access means refers to the memory means 15. The ATM interface according to claim 14, wherein cells corresponding to each time slot are read from the buffer memory. The control means stores a first bitmap consisting of a plurality of bit positions corresponding to the time slots for storing an empty state of the relative cell transmission time for each group, and an empty state of the cell transmission time. A second bitmap consisting of a plurality of bit positions corresponding to the time slots of
When the relative cell transmission time conflicts, an empty bit existing after the bit position corresponding to the relative cell transmission time is searched for by referring to the first bitmap corresponding to the group to which the cell belongs, and Is determined as the relative cell transmission time, and when the cell transmission times conflict, the empty bit existing after the bit position corresponding to the cell transmission time is searched with reference to the second bitmap, 16. The ATM interface according to claim 13, wherein a time corresponding to the bit is obtained as a cell transmission time. The control means includes register means for storing a plurality of bits in the first bitmap or the second bitmap as one block and storing a state of each block, and referring to the register means, the first bit 17. The ATM interface according to claim 16, wherein a reference range in the map or the second bitmap is specified. An ATM interface having a shaping function for controlling a transmission interval of an ATM cell to an output line,
A buffer memory for temporarily storing a plurality of ATM cells input from an input line;
Control means for writing cells to the buffer memory and reading cells from the buffer memory to the output line, wherein the control means
First table means for storing control parameters obtained corresponding to traffic conditions previously declared for each group and subgroup to which the input cell belongs;
Second table means for storing an empty state corresponding to a time slot on the output line;
When a cell arrives from the input line, the transmission timing of the cell is determined based on the control parameter corresponding to the group and subgroup to which the cell belongs stored in the first table means, and the second table means And access means for determining an empty transmission time slot to be associated with the transmission timing and reading out the corresponding cell stored in the buffer memory to the output line at the timing of the transmission time slot. An ATM interface, comprising:

Images