JPH03214945A - Call connection controller in packet switching network - Google Patents

Abstract

PURPOSE:To warrant communication quality of a call to required quality by classifying call classes into 1st and 2nd kinds, deciding the call class of each call and applying packet transfer of a call of a 1st call class with precedence over the call of the 2nd call class. CONSTITUTION:A call class deciding section 1 of each terminal equipment C identifies first whether the call class is a high speed call or a low speed call, the packet of the high speed call is distributed to a high speed call buffer 2 and a packet of the low speed call is distributed to a low speed call buffer 3 and they are sent to a multiplexing line ML from an output IF section 5 via the buffers 2, 3. A priority control section 4 applies flow control so that the packet of the high speed call buffer 2 is transferred at a peak rate, thereby transferring the packet with higher priority than that of the packet of the low speed call buffer 3. Thus, the packet of the high speed call is not aborted and the communication quality is warranted.

Description

[Detailed Description of the Invention] [Configuration of the Invention] (Industrial Field of Application) The present invention has a fixed packet length (ATM: Asynchr
The present invention relates to a call connection control device in a packet-switched network in the onous Transfer Mode.

(Prior Art) As shown in FIG. 7, in a packet switching network that accommodates multiple calls using the same communication resource, when setting up a call, terminals C, C,... - makes a call connection request to the network N side, and the network determines whether the communication quality of all calls accommodated by the network can be guaranteed, and decides whether to set up the call. It has become. In such a packet-switched network, the communication quality that network N must guarantee for each call is mainly two things: packet discard rate and packet transfer delay. The network receives that terminal C from the terminal side.
Estimate the packet discard rate and packet transfer delay in network N from the communication characteristic parameters of After determining whether the call is correct or not, the call is set up. As a result, the efficiency of using communication resources can be increased without deteriorating the communication quality below a certain standard.

However, the conventional packet drop rate estimated by the network and the packet drop rate required of the network for each call are generally the so-called average packet drop rate over a long period of time, so it is difficult to obtain the expected communication quality. There are aspects that cannot be avoided.

In other words, the packet discard rate in a short time or the chronological packet discard characteristics (for example, consecutive packet discards) are not taken into account at all, and the packet discard rate in a short time or the chronological packet discard characteristics are not considered in the network. was not guaranteed. Therefore, if each call has burst characteristics, packets are not necessarily discarded randomly, but are discarded relatively continuously, and even if the packet discard rate for a long time satisfies the communication quality required for each call. Even if it is controlled to do so, it is not possible to control short-term packet discards (particularly continuous discards), so it is not always possible to guarantee sufficient communication quality for each call. especially,
As shown in Figure 7, when there are multiple call types, the number of consecutively discarded packets is the number of consecutively discarded packets when comparing the number of consecutively discarded low-speed calls and the number of consecutively discarded high-speed calls. A phenomenon occurs in which the number of packets transferred becomes extremely large (in a high-speed call, approximately the number of packets transferred during a period of time approximately equal to the duration of the burst portion of the call is continuously discarded).

This problem of continuous discarding of packets becomes a major problem regarding data transfer. For example, this is a particularly big problem in the case of precise image data.

(Problems to be Solved by the Invention) As stated in the previous section, in conventional packet switching networks, the packet discard rate and packet transfer delay are controlled to ensure the communication quality of calls accommodated by the network. However, it does not guarantee the packet discard rate in a short time or the time-series discard characteristics of packets (for example, continuous packet discard), so the expected communication quality cannot be obtained and it is difficult to say that it is sufficient. there were.

In particular, this is a serious problem for high-speed calls because the number of consecutively discarded packets is much larger for high-speed calls than for low-speed calls.

The present invention has been made in view of the problems of the prior art, and its purpose is to ensure communication quality depending on the type of call.

[Structure of the invention]

(Means for Solving the Problems) A call connection control method in a packet switching network according to the present invention as set forth in claim 1 is such that when a plurality of calls are roughly classified into two call types, first and second, It is determined which of the first call type and the second call type each of the plurality of calls belongs to, and based on the call type determination, the call of the first call type is assigned to the second call type. Packets are transferred with priority over calls.

The call connection control method in a packet-switched network according to the present invention recited in claim 2 provides a call connection control system for a packet-switched network in which the ratio between the maximum call transfer rate and the multiplex line speed is greater than a predetermined value (that is,
High-speed calls) are the first call type, other calls (that is, low-speed calls) are the second call type, and the sum of the maximum transfer speeds of the calls of the first call type is less than or equal to the transfer speed of the multiplex line. In this case, all packets generated by the first call type are transferred by allocating the maximum bandwidth to the first call type.

The call connection control method in a packet switching network according to the present invention as set forth in claim 3 is characterized in that, based on the call type determination, a call of a first call type is preferentially transferred over a call of a second call type. In the call connection control to be performed, when the requested call is included in the first call type, the maximum transfer rate of all the already connected calls of the first call type and the above requested call are determined. If the sum is less than the transfer rate of the multiplexed line, and when the call request for the call request call is accepted, the packet discard rate required by the already connected call of the second call type can be met. On the other hand, when the above-mentioned call request call is a call of the second call type, the call request call and the already connected call of the second call type are received. When the packet discard rate for the call request call satisfies the request packet discard rate for the call request call and the already connected call of the second call type, the call request for the call request call is accepted. It is.

The call connection control method in the packet switching network according to the present invention as set forth in claim 4 uses the maximum transfer rate of the call and the transfer of the multiplexed line as means for estimating the packet discard rate when a new call request call is accepted. When the ratio with the speed is equal to a predetermined value, and the time required to transfer one packet at the maximum transfer speed is T, the first person who made the call request
The probability fi(n) that a call of the first call type generates n packets within the time T is detected by the declaration of the call, and the probability that the call of the first call type, including the call request call, n within T time
Find the probability f (n) of sending out packets, and calculate the packet discard rate of the second call type when the communication band of (T-n) packets/T time can be used with this probability f (n). This is then used as the estimated value.

The call connection control method in a packet switching network according to the present invention as set forth in claim 5 uses the traffic characteristics of the call as an estimated packet discard rate to be compared with the packet discard rate requested by the call in determining whether or not to connect the call. The packet discard rate is determined from the maximum transfer rate and average transfer rate of a call, assuming that packets arriving in excess of the number of packets that can be transferred within a predetermined period of time are discarded.

The call connection control method in a packet switching network according to the present invention as set forth in claim 6 is such that, when the ratio between the maximum transfer rate of a call and the transfer rate of a multiplexed line is equal to a predetermined value, the maximum transfer rate is set. If the time required to transfer one packet is T, then the probability fi(n) that call 1 of the first call type will generate n packets within the time T is calculated by the declaration of the call, etc. Detection■, n packets are detected during the relevant time T for the entire call of the first call type.
・ is calculated from the probability of occurrence f (n), and the packet discard rate when it is assumed that the first call type generates n packets is the maximum transfer rate and average transfer rate of the call type 2 call. The packet discard rate is determined by calculating the average value from f (n).

(Function) According to the present invention as set forth in claim 1, the name type is the first and the second.
After classifying each call into call types and determining the call type for each call, calls of the first call type are transferred with priority over calls of the second call type. By setting a call whose communication quality should not deteriorate as the first call type, it is possible to guarantee the communication quality of the call to the required quality.

Further, according to the present invention as set forth in claim 2, call types are divided into high-speed calls and low-speed calls, and when the sum of the maximum transfer speeds of high-speed calls is less than the transfer speed of the multiplexed line, the high-speed call is By allocating a peak rate to a call, all packets generated by the call are transferred, so the communication quality of high-speed calls, which are prone to deterioration, is reliably guaranteed.

According to the present invention as set forth in claim 3, when further call connection is performed, when the call request for the call request call is accepted, the call is made if the packet discard rate required by the low-speed call can be satisfied. Since the system accepts requests for making requested calls, the communication quality of low-speed calls is also reliably guaranteed.

According to the present invention as set forth in claim 4, when estimating the packet discard rate when a new call request call is accepted, a band that is likely to be unused among the bands allocated to high-speed calls is estimated. Since the packet discard rate when it can be used by low-speed calls is determined and used as an estimated value, communication resources can be used effectively on the premise of guaranteeing the communication quality of high-speed calls.

That is, in the method described in claim 3, it is assumed that communication quality is estimated using the remaining band of the band allocated at the maximum transfer rate for low-speed calls, but the use of the band allocated at the maximum transfer rate is Since the rate is not necessarily 100%, there is a probability that there is a communication band that is allocated to high-speed calls but is not used. By performing transfer control to use such unused bands for low-speed calls, the communication quality for low-speed calls can be improved, and in addition, the utilization rate of communication resources can be improved accordingly.

According to the present invention as set forth in claim 5, when determining whether or not to connect a call, the maximum transfer rate and average transfer representing the traffic characteristics of the call are used as the estimated packet discard rate to be compared with the packet discard rate requested by the call. From the speed, we calculated the packet discard rate assuming that packets that arrive in excess of the number of packets that can be transferred within a predetermined period of time are discarded. The amount of calculation can be reduced.

In other words, in a packet-switched network, in order to guarantee constant communication quality for each call because traffic fluctuates over time, it is necessary to use the communication resources of the packet-switched network and the current traffic state when setting up a new call. It is necessary to predict the communication quality when a new call is set up and to determine whether or not to set up a new call. As a method for this purpose, there is a conventional method of evaluating the packet discard rate based on the burst overflow rate (IEICE 1989 Fall National Conference B-204), but in the case of the burst overflow rate, the packet discard rate is evaluated based on the burst overflow rate. Assuming that when the sum of the maximum speeds exceeds the multiplexed line capacity, the excess portion is discarded, and the probability of such an event occurring is calculated from the ratio of the maximum transfer speed and the average transfer speed of each call, and the packet discard rate is calculated. This is to estimate the

This method of estimating the packet discard rate based on the burst overflow rate has a practical problem because the amount of calculation required to estimate the packet discard rate increases significantly when the number of call types accommodated by the packet switching network increases. .

According to the fifth aspect of the present invention, the packet discard rate can be determined from the maximum transfer rate and the average transfer rate with a smaller amount of calculation that does not cause any practical problems.

First, the maximum transfer rate and average transfer rate of a call give the probability that the call will generate a packet. by this,
When multiple calls are connected, the probability that a certain number of packets will arrive within a certain period of time can be determined. Thereby, the expected value of the number of discarded packets within the certain period of time can be determined.

When a new call request occurs, the estimated value of the packet discard rate for the currently connected calls, the number of connected calls, the probability that a certain number of packets will arrive, and the call request are To obtain a new packet discard rate at that point from the maximum transfer rate and average transfer rate of calls made without the need to newly calculate the probability of all calls that a certain number of packets will arrive within a certain time. I can do it.

In the present invention as set forth in claim 6, calls of the first call type and calls of the second call type are separated, and a probability of generating n packets within time T is given for the call of the first call type 1. As a result, estimation accuracy can be improved, and communication resource usage efficiency can be improved.

(Example) Examples of the present invention will be described below with reference to the drawings.

FIG. 1 shows an example of a flowchart of a call connection algorithm according to the present invention.

The network first determines whether the call of the terminal that made the call connection request is a high-speed call or a low-speed call (step S
101). Here, a high-speed call is a call in which the ratio BR between the maximum transfer rate of the call and the transfer rate of the multiplexed line ML is larger than a predetermined value BRo (a call with BR>BRo), and a low-speed call is a call other than that (a call where BR≦B R o).

If it is a high-speed call, it is determined whether a communication band at the maximum transfer rate (beak rate) can be secured for the high-speed call requesting call connection from the remaining band (step S102), and if the band is If it can be secured, the remaining bandwidth is updated (step 8103), and call connection at the peak rate is permitted (step S104).

Conversely, when it is determined that the band cannot be secured, the call connection is rejected (step S105).

On the other hand, if the call requesting connection is a low-speed call,
The packet discard rate is estimated (step S106), and the estimated value is compared with the requested discard rate of the outgoing request call to determine whether or not this estimated value satisfies the requested discard rate (step S107). . As a result of this determination, if the request drop rate is satisfied, call connection is permitted (step 8108).

Conversely, if the requested discard rate is not satisfied, the call connection of the requested call is rejected (step Sil).
o).

Packets generated by a call whose call connection has been controlled in this manner are flow-controlled by a priority control device P as schematically shown in FIG.

This priority control device P includes a call type determination unit 1 and a high-speed call buffer 2.
, a low-speed call buffer 3 , a priority control section 4 , and an output interface section (hereinafter referred to as an output IF section) 5 .

In each terminal C, a call type determination unit 1 first identifies whether the call type is a high-speed call or a low-speed call.

The call type is determined in this way, and conceptually, as shown in FIG. 6, high-speed call packets are distributed to high-speed call buffer 2, low-speed call packets to low-speed call buffer 3, and 3 from the output IF unit 5 to the multiplex line ML via
will be sent to.

The priority control unit 4 performs flow control so that packets in the high-speed call buffer 2 are transferred at a peak rate, so that they are transferred preferentially over packets in the low-speed call buffer 3. As a result, high-speed call packets are not discarded and communication quality is guaranteed.

Here, there are three examples of target bands for estimating the packet discard rate when performing call connection control for a group of low-speed calls.

■A method of estimating the packet discard rate using the bandwidth obtained by subtracting the bandwidth allocated to high-speed calls (beak rate) from the total bandwidth of the multiplexed line.

■A method of estimating the packet discard rate using the bandwidth obtained by subtracting the average bit rate of high-speed calls from the total bandwidth of the multiplexed line.

■A method for estimating the packet discard rate for low-speed calls from the overall bandwidth of the multiplexed line.

Next, an example of the packet discard rate will be shown.

(1) Method that takes statistical multiplexing effects into account ■ Method that uses a probability density function For low-speed calls, call connection control is performed using a probability density function that uses the number of packets generated for each call as a random variable. . The traffic characteristics of each low-speed call are expressed as a probability density function with the number of packets generated within a certain period as a random variable, and these functions are used to calculate the characteristics of multiplexed traffic. This method estimates the packet discard rate and performs call connection control. Note that this system has already been patented by the applicant of the present patent application (patent application number: 1999 Patent Application No. 103100).

■Management method using peak rate and average rate For low-speed calls, the packet discard rate is estimated from the peak rate and average rate. In this case, the packet discard rate for low-speed calls is estimated from the bandwidth allocated to low-speed calls or the entire bandwidth.

(2) Peak rate allocation method A method that uses peak rates to allocate bandwidth even for low-speed calls. In this case, the end result is that all calls are allocated bandwidth using their peak rates, and call connection control is performed.

FIG. 2 is a flowchart showing a call connection control algorithm according to the present invention.

In the network N shown in FIG. 7, the packet discard rate of an input call is estimated and call connection control is performed. All input packets have a fixed length, and the time required to transfer one packet to the multiplex line is defined as a unit time. In the network N shown in FIG. 7, two types of calls can be identified and different transfer controls can be performed. Here, for 2fili type call types, the ratio between the maximum transfer rate and the multiplex line rate is set to a predetermined value BR, similar to the control shown in FIG.
A call with a value larger than o is considered to be a first call type, and other calls are considered to be a second call type. The first type of call (hereinafter referred to as a high-speed call)
If the sum of the transfer speeds of Call type calls (hereinafter referred to as low-speed calls)
It is assumed that the device has a function that can transfer the packets generated. This can be achieved, for example, by performing transfer control to always transfer the packets preferentially when there are high-speed call packets in the buffer. Alternatively, a communication band equal to the sum of the maximum transfer speeds of high-speed calls is fixedly allocated on the multiplexed line, and when there are no high-speed call packets in the allocated band, low-speed call packets are transferred. Transfer control may also be performed.

Now, when a call request is issued from terminal i, it is determined whether or not call connection is possible as described below. Note that when making a call request, the maximum transfer rate Mi at which terminal i sends packets is declared at the same time as the call request. Also, the ratio between the maximum transfer speed and the multiplex line speed is B
Let the time required for a terminal with R o to send one packet be T (-1/BRo), and at least for high-speed calls, the probability of sending n packets within time T is f
1 (n) must also be declared. The packet switching network may know this based on the value set when connecting the terminal to the packet switching network.

Alternatively, the packet switching network may estimate the maximum transfer rate and average transfer rate of the terminal.

First, in the initial state (state where there is no connected terminal), the probability f (n) of generating n packets (n≧1) within time T during the entire high-speed call is set to "0". (Step S201).

When terminal 5 issues a call request, its maximum transfer rate M and multiplex line rate C. Find the ratio BR, -M,/l C (step 5203), compare this with BRo0 (step S204), and determine the call type of the requested call as
If BR is larger than B R o , it is determined that the call is a high-speed call; otherwise, it is determined that the call is a low-speed call (step S204).

If the requested call is a high-speed call, it is first determined whether the sum of the maximum transfer speeds of high-speed call terminals exceeds the communication band C1 that is allowed to be used by high-speed calls (step S205). ).

In this case, the sum of the maximum transfer speeds of high-speed call terminals is the communication band C
If the number exceeds 1, the call request is rejected (step S211).

Otherwise, calculate the packet discard rate CLR' for the already connected low-speed call (step S206,
207). Therefore, first, the probability f of sending n packets within time T for the entire high-speed call including the call request call is
'(n) is determined (step S206). This can be obtained by taking the l convolution of fi(n) for each terminal. CLR' is the probability f
Packet discard skewer CLR (1-n) when communication band (1-n/T) packet/unit time can be used in
n/T) and give it as the average (step S207). Next, this CLR' is compared with the packet discard rate CLRo required by the call included in the low-speed call (step 8208), and if the rate is satisfied, the call request is accepted (S209); otherwise, the call request is accepted. is rejected (step S211).

Note that when the call request call is accepted, f (n) is updated to f' (n) (step S 2 1
0).

If the determination in step S204 is that the call is a low-speed call, the packet discard rate CLR' for the low-speed call is calculated (step S212), and the call request is accepted through steps 8208 to S210, or the call request is accepted through steps 8208 and S210.
211 and rejects the call request.

According to this embodiment, the packet discard rate CLR (1-n/T) for a low-speed call when a communication band of (1-n/T) packets/unit time is available is input within T time. By calculating the expected value of the discard rate assuming that when the number of packets exceeds (T-n), that number of packets will be discarded, it is possible to give an upper limit to the actual packet discard rate, and to ensure a secure call connection. It is something that can be done.

FIG. 3 is a flowchart showing an algorithm for call connection control according to the present invention as set forth in claim 5. Before explaining the algorithm, an overview of call connection control in this embodiment and a method for calculating the packet discard rate will be explained. That's it.

Here, the traffic characteristics of a call are expressed by the maximum transfer rate and average transfer rate of the call, and packets that arrive in excess of the number of packets that can be transferred within a predetermined period of time are discarded. Find the discard rate from the maximum transfer rate and average transfer rate above, and use this value as
It compares the packet discard rate required by the call to determine whether or not the call can be connected.

Here, too, the time T required to send one packet from an output line in a packet switching network that communicates using fixed-length packets is treated as a unit time and used to calculate the packet discard rate.

Then, the maximum transfer rate at which the terminal sends packets is v1
It is assumed that N1 calls (calls of call type i) with an average transfer rate of rrlt packets/unit time and an average transfer rate of rrlt packets/unit time are connected. It is assumed that the number of call types is L. The call type with the highest maximum transfer rate is called PJ1, and T-1/v
Consider the packet discard rate per unit time. In the call fJfi, only one packet arrives at time vl/vl -T at most, so a call that can simultaneously generate packets at time T among calls 'Fii' N, has a probability al
o at n s−[N t ” V + / V l]
book ([X] represents the largest integer not exceeding X), probability a
01+1-[Ns-v i / v t ]
It is sufficient to calculate the packet discard rate assuming that there is +1 line.

The probability a10' 810 is alo- [NI- vI/v, ] +1-Nia
11””N1” Vs/Vt [Nt”V
l゜vl/vl /vl] Therefore, if there is no buffer, the expected number of packets discarded in time T, IcLR, is expressed as f (n
・・・． nL) represents the expected value of the number of discarded packets when call type l can generate nt packets simultaneously in time T. Next, f(nt...nL) is f("゜゛゜・") =E pj(”l・゜゜゜・“,)
(j −T)1゜ L j, r where p (n −・−, nL> is the call}i1i
It represents the probability that j packets will arrive within time T when i can generate J l' ni packets simultaneously during time T.

This p (n..., nL) is j l゜ It is.

As mentioned above, when calculating the packet discard rate by determining the CLR, a value larger than the packet discard rate in the actual switch is given, so if call connection control is performed based on this value, sufficient communication quality can be achieved for the connected call. can be guaranteed.

However, according to equation (1) pj (nl,...n L
) The larger L is, the larger the number of combinations to be calculated becomes, which increases the amount of calculation.

Therefore, when performing call connection control, calculations are performed with the following modification.

Consider the case where a call of call type i makes a new call request (
Therefore, the number of calls of call type i becomes N11+1). n1-[N1''Vs
/Vt], and when [(N+1)●vi/vl]-nl, the expected value CLR' of the number of waste packets at time T when a call request is included is cLR'-aso 't cni)+att't (nt
+1) (2) al1″″1-a10 (4) It can be written as. Here, f (mouth i) is defined as. So f(n+) and f(nx + 1)
By memorizing the value of , CLR' can be calculated.

Alternatively, the expected value CLR of the number of discarded packets/notes when calling request calls are not included is CLR=a1of (nρ0a,,t (nl+1)
(5), so f(n) and f (nt
+ 1), and by storing only the value of clr, formula (5
). The values of f(n) and f (n s +1 1) can be found from (6). However, Pj(n.》
is, ko11 p(n)-Σ...Σ j j d, -O dL-0 b akdR pj(nt"dt,"・, nt.
"', nL+dL)kj, and store the value of Pj (n1) itself, or 1)j-aIop3 (nj) +a,,pj(nj+
1) Using the relationship P. Calculate and use from the value of .

J [(N +1) ・v /v ] 1 nt +
When i If becomes l, CLR' = a (n +1) + a'
Since f (nt +2)10 l ll and f (n s +2) can be found from f(n1+1) using equation (6), CLR' can also be found in this case.

(7) The packet discard rate can be determined from CLR' and the load, and this packet discard rate is used to control call connection.

As an example, the case of L-2 will be described. The call FJI is the maximum sending rate Vl. at which the terminal generates packets. N1 calls with an average sending speed m are connected, call type 2 has a maximum sending speed V, and N2 calls with an average sending speed m are connected. n −N n − [N2 ”
V211'2 /Vl], and the expected value CLR of the number of discarded packets within time T at this time is CLR=a f (nn) + a2, f
(nt.
N2+1)” V2/vt] If one n2, a2
o'102+1-(N2+1)・v2/vla ”1
82g 2l 《On the other hand, the expected value CLR' of the number of rejected *,<kets including outgoing request calls is CLR' - a f (n n ) + a21f
(nl, n2+1) <8)2Ql'2.

f (n n +1) −f (n,, n2) +
1' 2 m2/V2・(1−ΣpJ (nt − n2) }(
9) Therefore, from this relationship and the value of CLR', f(nn)
and the value of f(n , n +1) as 1'2
12 and substitute it into equation (8).

[(N + 1) ・v / v] When 1 n 2 + 1, at 2 2 1, CLR'ma2o' f (n,, n2+1) +
a2l' f (nl, n2+2), so the formula (
9) to find f (nt .n2+2)
R' can be calculated.

The algorithm in Fig. 3 is based on the number N of currently connected calls of call FIIi, the probability that j packets will arrive in time T (p.1o≦j≦T-1), the load J, and the load ρ. , and CLR, and calculates the above-mentioned packet discard rate CLR' to perform call connection control.

First, in the initial stage, the number of calls currently connected for the call FJi is Nl, and the probability that j packets arrive at time T is {P. IO≦j≦T−1}, j Load ρ and initial values of CLR are set (step S301).

When a call request for call type l is detected (step S302), first, the probability P. (n)
-P. ([N,/BR1]) is calculated (Step S304), and based on this, the number of arriving packets f (n)-f ([・Ni/BR,]) and the number of packets arriving at the current time T are calculated. Call l Number of packets arriving at time T when including request call f (n
+1) = f ([N /BR+ 1 +1)
11 is calculated (step S304).

Then, [(N +1)/BR ] and [N1/1
1BR,] is equal (step S30
5).

If both are equal, according to equations (2), (3), and (4),
The expected value CLR' of the number of discarded packets at time T when the calling request call is included is determined (step S307).

In the determination in step S305, [(Nl +1)
/BR ] and [N /BR,] are not equal, that is, [(N +1)・BR]-01+11, then f (n +2)=f ([(N +1)/ B.R.
II +1) i l is determined (step 306), and CLR' is calculated by equation (7).
(Step S307).

An estimated value of the packet discard rate is obtained from the CLR' thus obtained, and this is compared with the packet discard rate required by the call to determine whether the call can be connected (step 308).

If the call connection is rejected, the process returns to step S302 and waits for a new call request.

If a call request is accepted, the CLR, N, Pj, which is a value representing the current state, is updated for later control (step S309), and the process returns to step S302.

From the above, the number of currently connected calls of call Mi is N, and the probability that j packets arrive at time T (P.
IO≦j≦T-1). Memorize the load ρ and J CLR, perform the calculations described above according to the algorithm shown in Figure 1, calculate CLR' when a new call request occurs, and perform call setting control. I can do it. In this case, when a call request occurs, the CLR
When calculating ', there is no need to calculate probabilities for all combinations, and the amount of calculation does not increase even if the number of call types increases. Further, by using the above-mentioned formula in the case of cancellation, the same amount of calculation can be performed.

FIG. 4 is a flowchart showing an algorithm for call connection control according to the present invention.

First, as a writing state, n(n≧1
) The probability f (n) of generating a call packet is set to "0°" (step S401).

When a call request is issued from terminal i (step S402)
The ratio BR between the maximum transfer speed M1 and the multiplex line speed C
, -Ml/Co (step 0 3403), compares this with the predetermined BRo (step S404), and sets the call type of the requested call to BR.
．． If is greater than BRo, then call type! 1, otherwise it is determined as call [2].

If the requested call is call N1, the requested call is T#
The probability of generating n packets in between f, (n) and f (
n), and the probability of generating n buckets in time T in the entire ^ speed call including the call request call is f'
(n) is obtained (step S405). Next, when it is assumed that n packets are generated from call type 1, the packet discard rate CLR(n) is calculated from the maximum transfer rate and average transfer rate of the already connected call FJ2, f' ( n
) to obtain the estimated value CLR' of the packet discard rate (step 5406). Here, CLR (n) can be determined by the algorithm shown in FIG.

The CLR′ obtained in this way is compared with the required packet discard rate CLRo (step S407), and the CLR
If ' is less than or equal to CLRo, accept the call request (
Step 3408), otherwise reject the call request (Step S411). When accepting a call request, replace f (n) with f' (n) (step S
409) Wait for next call request.

If the requested call is a call with call type 2, use the already saved value of f(n) to calculate the packet discard rate when a call with call type 1 generates n packets. Then, an estimated value CLR' of the packet discard rate is calculated (step S410). This CLR' is compared with CLRoS (step S407),
Determines whether the call request can be accepted.

Estimated value CLR' of the packet discard rate obtained in this way
gives an upper limit on the packet discard rate when each high-speed call has a buffer for the number of packets that can be generated within at most T time, and in such a system, it is impossible to make a secure call. This allows connection control.

〔Effect of the invention〕

As explained above, according to the present invention recited in claim 1, first, call types are classified into two types, first and second, and the call type is determined for each call. packet forwarding is given priority over calls of the second call type, so if you set a call for which you want to avoid deterioration of communication quality as the first call type, that call It becomes possible to guarantee the communication quality of the required quality.

Further, according to the present invention as set forth in claim 2, call types are divided into high-speed calls and low-speed calls, and when the sum of the maximum transfer speeds of high-speed calls is less than the transfer speed of the multiplexed line, the high-speed calls are By allocating a peak rate to a call, all packets generated by the call are transferred, so the communication quality of high-speed calls, which are prone to deterioration, is reliably guaranteed.

According to the present invention as set forth in claim 3, when further call connection is performed, when the call request for the call request call is accepted, the call is made if the packet discard rate required by the low-speed call can be satisfied. Since the system accepts requests for making requested calls, the communication quality of low-speed calls is also reliably guaranteed.

According to the present invention as set forth in claim 4, when estimating the packet discard rate when a new call request call is accepted, a band that is likely to be unused among the bands allocated to high-speed calls is estimated. Since the packet discard rate when it can be used by low-speed calls is determined and used as an estimated value, communication resources can be used effectively on the premise of guaranteeing the communication quality of high-speed calls.

According to the present invention as set forth in claim 5, when determining whether or not to connect a call, the maximum transfer rate and average transfer representing the traffic characteristics of the call are used as the estimated packet discard rate to be compared with the packet discard rate requested by the call. Based on the speed, the packet discard rate for already established calls, and the probability that the entire already established call will generate i packets within a predetermined period of time, Since the packet discard rate is calculated assuming that packets/notes arriving in excess of the available number of packets are discarded, it is possible to reduce the amount of calculation when a new call request is generated.

Further, in the present invention as set forth in claim 6, the first call type and the second call type are
By dividing calls of the first call type and giving the probability of generating n packets within time T for the call of the first call type, estimation accuracy can be improved and communication resource usage efficiency can be improved. can be raised.

[Brief explanation of drawings]

FIG. 1 is a flowchart showing a call connection control algorithm according to the present invention as set forth in claim 1.2, and FIG.
FIG. 3 is a flowchart showing a call connection control algorithm according to the invention according to claim 5, FIG. 4 is a flowchart showing a call connection control algorithm according to the invention according to claim 6. 5 is a block diagram of the priority transfer control system according to the present invention, FIG. 6 is a conceptual diagram of the buffer, and FIG. 7 is a flowchart showing the connection algorithm.
The figure is a schematic block diagram of a packet switching network to which the present invention is applied. S101...Call type determination step, S102 to S110
...Priority control step, S203.204...Call type determination step, S205~
212...Priority control step (S205-210...
- First control unit, S212, S208-S210... second control unit), S303-307... packet discard rate calculation step, S308... propriety determination step.

Claims (1)

[Claims] 1. When a plurality of calls are classified into two call types, a first call type and a second call type, each of the plurality of calls is classified into either the first call type or the second call type. call type determining means for determining whether the call type belongs to the call type, and a priority device for transferring packets to calls of the first call type more preferentially than calls of the second call type based on the output of the call type determining means. In a packet switching network comprising: a control means, when a request call is a call of the first call type, all calls of the first call type that have already been connected and the call originating from the first call type; When the sum of the maximum transfer rate with the call request call is less than the transfer rate of the multiplexed line, and the call request is accepted, the packet discard rate required by the already connected call type 2 call is satisfied. a first call request accepting means that accepts a call request for the call request call when the call request call is a call of a second call type; When the packet discard rate for the connected call of the second call type satisfies the requested packet discard rate for the call request call and the already connected call of the second call type, the call is made. A call connection control method in a packet switching network, comprising: second call request receiving means for accepting a request call request. 2. A call in which the ratio between the maximum call transfer speed and the multiplex line speed is larger than a predetermined value is considered to be the first call type, and the others are considered to be the second call type, and the priority control means If the sum of the maximum transfer speeds of calls of one call type is less than or equal to the transfer speed of the multiplexed line, the maximum transfer speed is assigned to the calls of the first call type. 2. The call connection control in a packet switching network according to claim 1, wherein all generated packets are transferred to give priority to calls of the first call type over calls of the second call type. method. 3. A call request is made when the ratio of the maximum transfer rate of the call and the transfer rate of the multiplexed line is larger than a predetermined value as the first call type, and other calls as the second call type. a call type determination means for determining to which call type a call belongs; and a call of the first call type is prioritized over a call of the second call type based on an output of the call type determination means. priority control means for transferring packets to the first call type, and the priority control means is configured to: when the call request call is included in the first call type, all the already connected first call types; If the sum of the maximum transfer speeds of all calls and the call request call is less than or equal to the transfer speed of the multiplex line, and the call request of the call request call is accepted, the already connected second a first call request accepting means that accepts a call request of the call request call when a packet discard rate required by the call type of the call can be satisfied; , the packet discard rate for the call request call and the already connected second call type is the same as the packet discard rate for the call request call and the already connected second call type. A call connection control device in a packet switching network, comprising: a second call request receiving unit that accepts a call request for the call request when the request packet discard rate with the call satisfies the request packet discard rate. 4. The first and second call request accepting means include means for estimating a packet discard rate when accepting a new call request call, and the packet discard rate estimating means includes: a maximum transfer rate of the call; If the ratio to the transfer rate of the multiplex line is equal to a predetermined value, and the time required to transfer one packet at the maximum transfer rate is T, then the first means for detecting a probability f_i(n) that a call of the call type generates n packets within the time period T;
means for determining a probability f(n) of transmitting n packets within a time; and a packet of a second call type assuming that a communication band of (T-n) packets/T time can be used with the probability f(n). 4. The call connection control method in a packet switching network according to claim 3, further comprising means for determining a discard rate. 5. Packet discard when the traffic characteristics of a call are expressed by the maximum transfer rate and average transfer rate of the call, and packets that arrive in excess of the number of packets that can be transferred within a predetermined period of time are discarded. The packet discard rate is calculated from the set of the maximum transfer rate, the average transfer rate, the packet discard rate for the already established calls, and the probability Pi that all the already established calls will generate i packets in the certain period of time. a packet discard rate calculation means; and a propriety determination means for comparing the packet discard rate calculated by the packet discard rate calculation means and the packet discard rate required by the call to determine whether call connection is possible or not. A call connection control method in a switched network. 6. Calls in which the ratio between the maximum transfer rate and the multiplex line speed is larger than a predetermined value are considered to be the first call type, and the others are the second call type, and the maximum transfer rate and the multiplex line rate are Let T be the time required to transfer one packet at the maximum transfer rate when the ratio to the rate is equal to the predetermined value, and for the first call type call i, n within the T time. means for detecting the probability f_i(n) of generating n packets within the time T for the entire call of the first call type; ), and the packet discard rate when it is assumed that n packets are generated from the first call type is determined from the maximum transfer rate and average transfer rate of the low-speed call, and with the probability f(n), 1. A call connection control method in a packet switching network, comprising a packet discard rate estimating means for calculating an average and estimating a packet discard rate.