JPH04280532A - Call connection control system in packet switching network - Google Patents

Abstract

PURPOSE:To reduce the calculation quantity by calculating maximum transfer speed and average transfer speed while a packet abort rate is requesting a call a probability of production of a packet in a required time by the entire calls set already, and the number of a call type at the same characteristic of a call making a call request. CONSTITUTION:One basic time is defined to be a time when a multiplex line or a switch transfers one fixed length packet. Let a maximum transfer speed of a call having a largest transfer speed among multiplexed calls be Vso and T be 1/Vso, then T sets of packets are transferred for a time T through a multiplex line. When the sum of number of packets generated in each call for the time T exceeds T, packets in excess of the T are aborted. Moreover, an abort rate larger than the actual packet abort rate is estimated by adopting the packet abort model in this way and the system operation is implemented at the safety side.

Description

[Detailed description of the invention]

[Structure of the invention]

0002

[Industrial Application Field] The present invention has a fixed packet length (A
TM: Asynchronous Transfer
This invention relates to a call connection control method in a packet-switched network (Mode).

0003

2. Description of the Related Art Generally, as shown in FIG. 6, in an ATM packet-switched network in which multiple types of calls are accommodated using the same communication resource, when setting up a call, the terminal side makes a call connection request to (step S1),
The network estimates the communication quality in the network from the communication characteristic parameters received from the terminal side (step S
2) After determining whether the communication quality of all calls accommodated by the network, mainly the packet drop rate and packet transfer delay, can be guaranteed (step S3), call setup (step S4) or call setup is performed. is rejected (step S5). This increases the efficiency of using communication resources without deteriorating the communication quality below a certain standard.

0004

[Problems to be Solved by the Invention] However, in the conventional call connection control system in the ATM network described above, multiple types of calls with different communication characteristics are accommodated, so when multiple call types are accommodated, each call It becomes necessary to estimate communication quality. Therefore, a means to estimate the communication quality is required, and the amount of calculation required to solve the problem of multiple types of calls sharing a single communication resource increases exponentially with the number of call types. do. Therefore, there is a problem in that it takes time to determine whether a call can be connected, and real-time performance is degraded.

[0005] In view of the above-mentioned problems, the object of the present invention is to
This invention provides a call connection control method in a packet switching network that reduces the amount of calculation and improves real-time performance. [Structure of the invention]

[0006]

[Means for Solving the Problems] In order to achieve the above-mentioned object, the present invention provides a packet discard rate when packets arriving in excess of the number of packets transferred within a predetermined time are discarded. are the maximum transfer rate and average transfer rate of the call making the call request, the probability that all the calls that have already been set up will generate packets in a predetermined time, and the type of call with the same characteristics as the call making the call request. The packet discard rate calculated above is compared with the packet discard rate required by the call to determine whether call connection is possible.

[0007]

[Operation] In the present invention, the packet discard rate is calculated based on the maximum transfer rate, the average transfer rate, the probability that all already set calls will generate packets in a predetermined time, and the call request. Calculated from the number of call types with the same characteristics as the call being made. In other words, since weighted averaging is performed when calculating the packet arrival probability, the amount of calculation required to estimate the packet discard rate is reduced, and the amount of memory required for estimation calculation is also reduced.

[0008]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the call connection control method of the present invention will be described with reference to FIGS. 1 to 5.

FIG. 1 is a flowchart showing a control algorithm for a basic estimation method of packet discard rate. According to the figure, one basic time is defined as the time required for a multiplexed line or switch to transfer one fixed-length packet, and one basic time is defined as the time required for a multiplexed line or switch to transfer one fixed-length packet. Let the maximum transfer rate of Vm0 (packet/basic time) be Vm0 (packet/basic time). Furthermore, if T is defined as 1/Vm0, the multiplex line or switch has the ability to transfer T packets during time T. On the other hand, each call transfers only one packet during time T.

The packet discard rate will be estimated on the assumption that packet discards occur as follows. When the total number of packets generated from each call in time T exceeds T, the more than T packets are discarded. Furthermore, by adopting such a packet discard model, we will estimate a higher packet discard rate than the actual packet discard rate.
You will be on the safe side.

[0011] L types of calls are multiplexed, and the number of calls of call type i is Ni (maximum transfer rate; Vmi, average transfer rate; Va
i).

If Pk is the probability that k packets will arrive from each call within time T, then the packet discard rate is given by the following equation (1). However, ρ is the load factor.

[0013]

[Math 1]

At this time, assuming that a call of call type i is newly set up, the packet discard rate is calculated as follows using {Pk} for the currently set call.

First, a call processing request (step S2) is made from the initial state (step S1), and a call request (step S2) is made.
In the case of 3), the ratio αi between Vmi and Vai is calculated from the following equation (2).
and ρ' are obtained (step S4).

[0016]

[Math 2]

Then, {Pk'} is obtained from the following equation (3) (step S5).

[0018]

[Math 3]

Furthermore, CLR' is obtained from the following equation (4) (step S6).

[0020]

[Math 4]

[0021] Here, CLR' and CLRo (threshold) are compared (step S7), and CLR' is CLRo.
If it is smaller than that, an acceptance process is performed (step S8).
, ρ=ρ', {Pk }={Pk '} (Step S9
) and returns to step S2. Also, CLR' is CLR
If it is larger than o, perform rejection processing (step S10).
, return to step S2.

Similarly, when a call of call type j cancels, αj and ρ' are calculated from the following equations (5) and (6) (step S1
1),

[0023]

[Math 5]

[0024]

[Math 6]

[0025]

[Math 7]

{Pk'} is obtained from the above equation (7) (step S12). Then, a cancellation process is performed (step S1
3), return to step S2.

[0027] In this way, by introducing the rounded equation to the equation for estimating the basic packet discard rate, the amount of calculation is reduced. The new packet discard rate after a new call is set up is determined by the packet arrival probability before the new call is set up (up to T arrival probability is sufficient), the current load factor, and the newly set packet arrival probability. Calculated from the maximum call transfer rate/average transfer rate. Furthermore, this estimation method does not depend on the number of call types L at all, and the amount of calculation does not increase at all even if the number of call types increases. Further, the amount of calculation in this estimation method is of order T.

However, in this basic estimation method, it is assumed that each call can transfer packets at Vm0, which is the maximum transfer rate of the fastest call among all calls (the average transfer rate is maintained), and the packets are discarded. Since this method estimates the packet drop rate, it has the drawback of estimating a packet discard rate that is much higher than the actual rate.

FIG. 2 is a flowchart showing a control algorithm of an estimation method for converting a low-speed call into a high-speed call in order to eliminate such drawbacks.

In the figure, it is sufficient to consider that Ni calls of call type i can transfer only Ni·T/Ti packets within time T (Ti=1/V). On the other hand, a call with a maximum transfer rate of V (call type 0) can transfer a maximum of one packet within time T. Therefore, Ni calls (call type i) are
This can be regarded as T/Ti calls of type 0. In this way, by converting a plurality of low-speed calls into high-speed calls, it is possible to take into account in analysis that packets cannot be transferred by low-speed calls at the maximum transfer speed of high-speed calls.

However, since Ni·T/Ti is not necessarily an integer, it is possible to compensate by the following method depending on the case. In other words, the drop rate when there are Ni・T/Ti calls is the number of calls of call type i between [Ni・T/Ti] and [N
i·T/Ti]+1 is used for compensation. However, [x] is the largest integer that does not exceed x.

That is, a call processing request is made (step S2) from the initial state (step S1), and the following equations (8) and (9)
Then, weights ai0 and ai1 are obtained (step S3).

[0033]

[Math. 8]

[0034]

[Math. 9]

The packet discard rate when a new call is set up can be found recursively as follows.

If the network side knows the probability of arrival of k packets {Pk}, then ni = [Ni・T
/Ti], the probability that k packets arrive when there are ni calls of call type i {Rk (i, ni)
}, {Rk (i, ni +1)} as shown in the following equation (10), (
11) at any time (step S4). At this time, {Rk (i, ni)} and {Rk (i, n
i +1)} does not need to be memorized. Note that it is sufficient for {Pk }, {Rk (i, ni)} and {Rk (i, ni +1)} to be stored and determined to have probabilities in the range 0≦k≦T.

[0037]

[Math. 10]

[0038]

[Math. 11]

The packet discard rate estimation when a new call of call type i is set (step S5) is [(Ni+1)・T
/Ti] is equal to ni, the above equations (10) and (11
) using {Rk (i, ni)}, {Rk (i,
ni +1)}. [(Ni+1)・T/Ti]
When increases from ni to ni +1 (step S6), the following equation (12) is used to calculate {Rk (i, ni +2
)} is determined (step S7).

[0040]

[Math. 12]

Then, the new packet arrival probability {Pk
'} is obtained from the following equation (13) (step S8),

00
42]

[Math. 13]

CLR' is obtained from the following equations (14) and (15) (step S9).

[0044]

[Math. 14]

[0045]

[Math. 15]

Furthermore, CLR' and CLRo are compared (step S10), and if CLR' is smaller than CLRo, an acceptance process is performed (step S11), and Ni=
Ni+1, ρ=ρ', {Pk}={Pk'} (step S12), and the process returns to step S2. Also, CLR'
When is larger than CLRo, a rejection process is performed (step S14), and the process returns to step S2.

Furthermore, in step S6, [(Ni
+1)·T/Ti]=ni, obtain {Pk'} from the following equation (16) (step S13).

[0048]

[Math. 16]

Similarly, in the case of cancellation, [(Ni-1)・
T/Ti]<ni (step S15), {Rk (i, ni -1)} is obtained from the above equation (12) (step S16), and {Pk '} is obtained from the above equation (13). (step S17), and performs cancellation processing (step S1).
8). Further, in step S15, [(Ni-1)・T/
Ti]=ni, the process advances to step S17.

FIG. 3 is a flowchart showing a control algorithm for a method of estimating the packet discard rate by distinguishing between high-speed calls and low-speed calls.

In the figure, first, a call processing request is made (step S2) from an initial state (step S1), and the following equation (17) is performed for a low-speed call and a high-speed call for a predetermined time T.
Define it like this.

[0052]

[Math. 17]

Furthermore, the probability {F(n)} that a high-speed call transfers n packets within time T is calculated using the following equations (18) and (19).
). At this time, assume that high-speed calls maintain the average transfer rate, but when transferring packets, packets are transferred continuously at the maximum transfer rate V for a time T, and F(n)
Calculate. This results in evaluating the worst-case packet arrival pattern in terms of packet drop rate.

[0054]

[Math. 18]

[0055]

[Math. 19]

[0056] However, <x> is the smallest integer greater than or equal to x.

[0057] Then, it is determined whether the call is a high-speed call or a low-speed call (step S3), and a new high-speed call (call type j) is set (
Step S4) Sometimes, by the following decompression formula (20), {
F'(n) } can be easily obtained (step S5).

[0058]

[Math. 20]

When the number of packets transferred for a high-speed call is n, it is possible to transfer (T-n) packets for a low-speed call. Since a total of more than T packets are discarded, the packet discard rate CLR, which is the sum of high-speed calls and low-speed calls, can be obtained from the following equations (21), (22), and (23) (step S6). .

[0060] However, let ρL be the average load factor of low-speed calls, and let Pk be the probability that the total number of packets arriving from the low-speed call(s) is k.

[0061]

[Math. 21]

[0062]

[Math. 22]

[0063]

[Math. 23]

Furthermore, comparing CLR and CLRo (
Step S7), and when CLR is smaller than CLRo, the reception process is performed (step S8), and {F(n)}=
{F'(n)} (step S9), step S
Return to 2. Also, for cancellation, {F'(n)
} are gradually obtained (step S10), and a cancellation process is performed (step S11).

[0065]

[Math. 24]

Conversely, in step S7, CLR is
When it is larger than LRo, rejection processing is performed (step S
12), return to step S2.

On the other hand, if the call for which the call processing request is made is a low-speed call and is a calling request (step S13), the above equation (
3) or (13) to obtain {Pk'} (step S14). Then, calculate the CLR from the above equation (23) (
Step S15). Here, CLR and CLRo are compared (step S16), and if CLR is smaller than CLRo, an acceptance process is performed (step S17), and {P
k }={Pk'} (step S18), and the process returns to step S2. Further, when CLR is larger than CLRo, a rejection process is performed (step S21), and step S2
Return to

Similarly, when canceling the call, {Pk'} is obtained from the above equation (7) or (13) (step S19), and the canceling process is performed (step S20).

FIG. 4 is a flowchart showing a control algorithm for estimating the packet discard rate when high-speed calls are preferentially transferred.

In the estimation method shown in FIG. 3, assuming that n packets transferred by a high-speed call are preferentially transferred without being discarded, and that the high-speed call transfers n packets,
The number of packets remaining in the low-speed call that can be transferred is Tn packets. At this time, the packet discard rate CL of the low-speed call when the low-speed call can output T-n packets within T time
R(n) is obtained from the following equation (25). However, the sum of the maximum transfer rates for high-speed calls shall not exceed the transfer rate of the multiplexed line or the packet processing capacity of the switch.

[0071]

[Math. 25]

At this time, in steps S6 and S15, the total packet discard rate CLR is estimated by the following equation (26). In addition, if the sum of the maximum transfer speeds of high-speed calls exceeds the transfer speed of the multiplex line or the packet processing capacity of the switch, CLR(n) can be calculated using the above formula (25).
and is determined by the following equation (27). Here, when the sum of the maximum transfer speeds of high-speed calls is less than or equal to the transfer speed of the multiplexed line, the above equation (25) is used, and when it exceeds the sum, the following equation (27) is used. The total packet discard rate CLR is estimated by the following equation (28).

[0073]

[Math. 26]

[0074]

[Math. 27]

[0075]

[Math. 28]

Note that the other steps are the same as those in FIG. 3, so their explanation will be omitted here.

FIG. 5 shows an evaluation example of this embodiment. The solid line is the estimated value, and the other three lines are the simulation results. Two types of calls were multiplexed to find a load that would result in a packet discard rate of 10-5 for low-speed calls. This shows that estimation is performed on the safe side and with relatively high accuracy.

[0078]

[Effects of the Invention] As explained above, according to the present invention, weighted averaging is performed when calculating the packet arrival probability, so the amount of calculation required for estimating the packet discard rate can be reduced.
The amount of memory required for estimation calculations can also be reduced. Therefore, the call connection control method can be realized with a small amount of calculation. Furthermore, responsiveness to call connection requests is improved, and real-time performance can be improved.

[Brief explanation of the drawing]

FIG. 1 is a flowchart showing a control algorithm for a basic estimation method of the present invention.

FIG. 2 is a flowchart showing a control algorithm of an estimation method for converting a low-speed call into a high-speed call according to the present invention.

FIG. 3 is a flowchart showing a control algorithm of a method of distinguishing between high-speed calls and low-speed calls and estimating a packet discard rate according to the present invention.

FIG. 4 is a flowchart showing a control algorithm of an estimation method when priority transfer is performed on high-speed calls according to the present invention.

FIG. 5 is a diagram showing an evaluation example of the present invention.

FIG. 6 is a flowchart showing a conventional call connection control procedure.

Claims (1)

[Claims] Claim 1: The packet discard rate when packets arriving in excess of the number of packets transferred within a predetermined time period are discarded is defined as the maximum transfer rate of the call making the call request; Calculated from the average transfer rate, the probability that all the calls that have already been set up will generate packets in a given time, and the number of call types with the same characteristics as the call making the call request, and the packet discard rate calculated above and the call A call connection control method in a packet-switched network, characterized in that a decision is made as to whether or not a call can be made by comparing the packet discard rate requested by a packet discard rate.