RU2547631C2 - Method for efficient use of communication resources of multiservice network in congestion conditions - Google Patents

Abstract

FIELD: physics, computer engineering.

SUBSTANCE: invention relates to multiservice communication networks. The method includes setting a table of priorities, wherein each user terminal is assigned a priority at its static IP address; determining the priority of a new connection from the table of priorities in accordance with the IP address of the sender terminal; reading values of the required resource for the new connection from the header of its IP address; calculating free access network resources; determining if there are existing open connections with a lower priority than the new connection; comparing their overall resource with the required resource for the new connection; if insufficient, the connection is denied service; if sufficient, the relative resource consumption of said connections is calculated; ordering the selected open connections; selecting from the ordered open connections one or more connections, starting with the connection with the maximum value, after which said connections are terminated, and the freed resource is provided to the new connection.

EFFECT: high efficiency of using communication resources in multiservice networks.

1 cl, 3 dwg

Description

Technical field

The invention relates to the field of digital information transmission by data exchange in packet-switched systems, namely to multiservice communication networks.

State of the art

Multiservice network (multi-service network) - a multifunctional network with an unlimited range of services and flexible capabilities for their management, personalization and the creation of new services through the unification of network solutions [V.V. Velichko, E.A. Subbotin, V.P. Shuvalov, A.F. Yaroslavtsev. "Telecommunication systems and networks", "Hot line - Telecom", 2005, p.580].

It is known that the quality of service in multiservice networks (MSS) is ensured by allocating network resources individually to each individual data stream at the network level (microflow) (hereinafter - to the connection). The allocation of network resources at the connection level to ensure quality of service is provided in the access network, in the network segment that supports the IntServ architecture [Braden R., Clark D., Shenker S. "Integrated Services in the Internet Architecture: Overview", RFC 1633, June 1994 ]. In order to prevent and eliminate network congestion when establishing access to network resources, and ensuring the coordination of network parameters and user traffic in switching nodes, Admission Control systems are used. Access control is explicitly used in the switching nodes of the IntServ architecture along with the mechanism of preliminary reservation of resources in the switching nodes for individual connections. As resources, it is primarily considered the bandwidth of communication channels. In IP-oriented networks, the most common redundancy mechanism is the RSVP protocol [Braden R., Zhang L., Berson S., Jamin S. "Resource Reservation Protocol (RSVP) - Version 1 Functional Specification" RFC 2205, September 1997 ]. The input of the system receives information about the TSpec stream descriptor and the set of parameters of the required RSpec service [S. Shenker, C. Partride, R. "Guerin. Specification of Guaranteed Quality of Serrvice". September 1997. RFC 2212]. Among these parameters, peak transmission rate, average transmission rate, maximum allowable delay, probability of packet loss, maximum packet size are specified. One of the methods for transmitting the values of the listed parameters is “The method, system and communication device for informing and providing the parameters of the quality of service profile in the network” (RU patent No. 2337492 dated 10.27.2008, bull. No. 30, IPC H04Q 7/38). Based on these values of the flow parameters and the requirements for the quality of service, the required amount of resources is calculated that is sufficient to service this connection request. In this case, the throughput can be allocated according to average, peak values or calculated on the basis of one of the many flow models. The most fully known access control methods are presented in [Kucheryavy EA Traffic management and quality of service on the Internet. - St. Petersburg: Science and Technology, 2004 - 336 pp.], In which the decision on admission of the application is made either on the basis of the analysis of the parameters of the new flow (parametric admission control) [Kraimeche V., Schwartz M. "Analysis of traffic access control strategies in integrated services networks ", IEEE Transactions on Communications, vol. 33, no.10, pp.1085-1093, October 1985], or based on the results of measuring the state of the network infrastructure engaged in servicing accepted requests (admission control based on measurements) . In the general case, in the aforementioned methods, upon receipt of an application for data transfer, the access control task consists in solving two issues: are there free requested resources on the network; whether the new load will affect the quality of service of already existing flows and connections, on the basis of which a decision is made on the admission or not admission of a new connection for service.

There is a known method of “Limiting traffic for a network with transmission with levels of quality of service” (patent RU 2299516 C2, publ. 05.20.2007, bull. No. 14), which checks the admissibility for the transfer of a group of data packets with priority, and in the case of a positive decision data flow to service; if the validation checks are negative, the group of data packets is rejected or transmitted with a lower priority or without priority.

The disadvantage of this method is the prioritization of connections only according to classes of quality of service in accordance with the model of differentiated services and does not take into account priority for user categories. There is no possibility of redistributing network resources for priority connections in conditions of congestion due to already served connections.

There is also a known method of servicing priority user services with dynamic allocation of network resources, including under conditions of network congestion "Reliable packet data services associated with communication networks" (patent RU 2461997 C2, published on September 20, 2012, Bulletin No. 26). The disadvantage of this method is the inability to take into account the resource consumption of incoming connections when accessing the network, as well as the procedure for selecting terminated connections of lower priority during network congestion.

The closest in technical essence and the functions performed to the claimed one is "Method and device for the efficient use of communication resources in a data exchange system under overload conditions" (patent RU 2286649, C2 H04L 12/56, published on 10.27.2006), according to which a request is detected to open a new connection between the access terminal and the data network for data transmission, it is determined whether the access network between the access terminal and the data network is in an overloaded state in response to a detected request, if the access network but it is in a state of overload, it is determined whether there are any already existing idle open connections in the access network between the access terminal and the data network, while each idle open connection has communication resources assigned to it, one of the already mentioned idle open connections is selected on based on at least one of (a) open connection times and (b) previous data traffic activity of said already existing unoccupied open connections, the mentioned selection is broken This already existing unoccupied open connection, communication resources corresponding to said torn selected already existing unoccupied open connection are allocated to said new connection.

The disadvantages of the prototype method include the lack of connection priorities and the accounting for the resource consumption of open connections that are torn, which can lead, on the one hand, to rupture or not providing services to priority connections, on the other hand, situations when resources are released from one broken connection is not enough to service a new connection, which will require breaking more than one connection and will lead to inefficient use of telly communication resources.

Disclosure of invention

The objective of the invention is to develop a method for the efficient use of communication resources of a multiservice network under congestion conditions, which allows to obtain a more efficient use of resources of a multiservice network by taking into account connection priorities and resource consumption of disconnected connections.

This problem is solved in that the method of efficient use of the communication resources of a multiservice network in conditions of congestion, which consists in detecting a request to open a new connection between the access terminal and the data network, determines whether the access network between the access terminal and the data network is in congestion in response to a detected request, if the access network is in a state of congestion, then it is determined whether there are any existing idle open connections in the access network UPA between the access terminal and the data network, with each unoccupied open connection having communication resources assigned to it, the mentioned selected already existing unoccupied open connection is severed, the communication resources corresponding to the said severed selected already existing unoccupied open connection are allocated, additionally set the priority table in which each access terminal is prioritized by its static IP address. After a request to open a new connection is detected and network congestion is determined, the priority of the new connection is determined from the priority table in accordance with the IP address of the sender terminal. The priority value is the greater, the greater the privileges the user and the services provided to him have. Read the values of the required resource for the new connection from the header of its IP address. The idle resources of the access network are calculated. It is determined whether there are already existing open connections with lower priority than the new connection, their total resource is compared with the required for the new connection. If it is not enough, then it is denied service, if enough, then the relative resource intensity of these connections is calculated, which is the ratio of the allocated bandwidth of each connection to its priority value. Sort selected open connections by relative resource intensity from a larger value to a smaller one. One or more from ordered open connections is selected starting from the connection with the maximum value until the condition is fulfilled when the sum of the resources of the selected connections and unoccupied resources is sufficient to service the new connection. Then these connections are broken, and the freed resource is provided to the new connection.

The listed new set of essential features provides the ability to more efficiently use the communication resources of the multiservice network when servicing priority connections by taking into account connection priorities and resource consumption of disconnected connections. Therefore, the claimed invention meets the condition of patentability "Inventive step".

The analysis of the prior art made it possible to establish that analogues that are characterized by a combination of features that are identical to all the features of the claimed technical solution are absent, which indicates the compliance of the claimed method with the condition of patentability "novelty".

Brief Description of the Drawings

The claimed object of the invention is illustrated by drawings, which show:

figure 1 is a structural diagram of a switching node with an access control system;

figure 2 is a flowchart for the efficient use of network resources in an overload condition;

figure 3 is a graph of the probability of service failure on the incoming load on the network, taking into account the priorities of the connections.

The implementation of the invention

The implementation of the claimed method can be explained on the diagram of the switching node with the implementation of the admission control of connections to the network, shown in figure 1. An application with RSVP - 2 functions located in the user terminal - 1 receives an application for servicing a new connection n + 1 to the input port of the Intserv-3 network switching node under consideration, where a static IP address is read in the packet classification block - 6 and transmitted to the block Priority determination - 5, which is structurally located in the network access admission control unit - 4, where, based on the priority table, the priority of the new connection is determined, which is subsequently transferred to the admission and redistribution unit of resources - 7. Also, in the packet classification block - 6, Tspec parameters are read based on the peak or average values of the bandwidth of the new connection and transferred to the block for determining the required resources for the new stream - 8, which then go to the block for admission and redistribution of resources - 7. In addition, from the block management of local resources of the packet scheduler - 10 data about all reserved bandwidths for already served connections is transmitted to the block for determining unoccupied resources - 9. In the block for determining unoccupied of resources, unallocated bandwidth is determined as the difference between the bandwidth of the output port of an access network element and the total bandwidth allocated to already served connections, and the received values are transferred to the resource admission and redistribution block - 7. In the resource admission and redistribution block - 7, the required for the requested new connection bandwidth with idle, if unoccupied resources are more than requested, then the new resource is allocated the required resource, If unoccupied resources is less than requested, it is determined whether there are existing open connections w _i with a priority lower than the requested compound of w _{n + 1,} and if they are not or less required total resource requests a connection, he denied service. If there are open connections with lower priority and their total resource is not less than required by the requested connection, then the relative resource consumption of these connections is calculated. The relative resource consumption of open connections is calculated as the ratio of the allocated bandwidth to each connection to its priority. One or more of the existing open connections is selected, starting from connections with the maximum relative resource intensity until the total resource of the selected connections is sufficient to service the new connection. When resources are sufficient, information about one or more existing open connections selected for breaking existing connections and allocation of the required bandwidth to the new connection are transferred to the local resources control unit of the packet scheduler.

Figure 2 shows a flowchart for the efficient use of resources of a multiservice network in an overload environment. At step 13, a request to open a new connection between the access terminal and the data network is detected. At step 14, the traffic parameters of the requested connection are read from the fields of the first packet and its priority is determined, the required resources for the requested connection are determined. At step 15, unallocated resources are calculated by the formula

, $$R_{nes\mathrm{but}n}=\mathrm{FROM}-{\displaystyle \sum _{i=\mathrm{one}}^n{r}_i}$$

,

where C is the bandwidth of the access network element;

- суммарная пропускная способностью открытых соединений. $${\displaystyle \sum _{i=\mathrm{one}}^n{r}_i}$$

- total throughput of open connections.

At step 16, the throughput required for the requested connection is compared with unoccupied, if the unoccupied throughput is not less than requested, then at step 17 the required resources are allocated to this requested connection. If the unallocated bandwidth resources are less than the requested resources, then at step 18 it is determined whether there are already existing open connections with a lower priority than the requested connection. At step 19, check for open connections with lower priority than the requested connection, and also verify that the total throughput of open connections with priority less than the requested connection is less than the required bandwidth for the requested connection. At step 20, if the above conditions are not met, service is denied. If there are open connections with a lower priority, then at step 21, the relative resource intensity of these compounds is calculated by the formula

, $$g_i=\frac{r_i}{W_i}$$

,

where r _i is the allocated bandwidth of the i-th connection;

w _i is the priority of the i-th connection.

At step 22, the selected open connections are ordered by relative resource intensity from a larger value to a smaller one. At step 23, one or more of the existing open connections is selected, starting from the connection with the maximum relative resource intensity, until at step 24, the sum of the resources of the selected connections and unallocated resources is not sufficient to service the new connection. At step 25, if there are sufficient resources, one or more selected pre-existing open connections are broken. At step 26, released communication resources are allocated to the new connection.

The ability to achieve the claimed technical result was tested by machine simulation with well-known software (VINT project network simulator version 2 (NS2)).

Streams from 160 sources of three priorities were used as a load, each of which randomly generated a load with peak throughput values of 1400 bit / s, 87 bit / s, 64 bit / s. Transmission time with a maximum speed of 1s, 3s, 5s, respectively. The maximum packet size is 900 bytes. The number of load sources of the first priority was 30, the number of sources of the second priority was 60, and the number of sources of the third priority was 70. The bandwidth of the port of the switching node is 100 Mbps, the queue length is unlimited.

The achievement of the technical result is illustrated as follows. For the prototype method, the required rate of loss of applications P _req = 0.7, determined in accordance with the standards specified in the main technical requirements, which are mandatory when designing switching systems [Norms of technological design. Urban and rural telephone networks. RD 45.120-2000. NTP 112-2000, Ministry of the Russian Federation for Communications and Informatization, 2000, p. 25], is achieved with a simulated load of 6.1 Earl. For the developed method, this norm is achieved at a load of 7.65 Earl. The difference between the network loads when using the prototype method and the developed method is 1.55 Earl, taking into account the priorities of the connections, which is what the formulated technical result is achieved when implementing the claimed method.

Thus, the claimed method, by taking into account the priorities of the connections and calculating their relative resource consumption while accessing the network resources, allows efficient use of communication resources when servicing priority connections in multiservice networks under congestion conditions.

Claims (1)

A method of efficiently utilizing the communication resources of a multiservice network under congestion conditions, which consists in detecting a request to open a new connection between the access terminal and the data network, determining whether the access network between the access terminal and the data network is in an overload state in response to the detected request if the access network is in an overloaded state, then it is determined whether there are any existing idle open connections in the access network between the access terminal and a data network, wherein each unoccupied open connection has communication resources assigned to it, the mentioned selected already existing unoccupied open connection is torn, the communication resources corresponding to the said torn selected already existing unoccupied open connection are allocated to the new connection, characterized in that they additionally define the table priority, in which each user terminal is assigned priority by its static IP address, after detection requests to open a new connection and determine network congestion determine the priority of the new connection according to the priority table in accordance with the IP address of the sender terminal, read the values of the required resource for the new connection from the IP address header, calculate unoccupied access network resources, determine if there are already existing open connections with lower priority than the new connection, compare their total resource with the required for the new connection, if it is not enough, then it is denied service, if enough, then the relative resource intensity of these compounds is calculated, which is the ratio of the allocated bandwidth of each connection to its priority value, the selected open connections are ordered by relative resource intensity from a larger value to a smaller one, one or more are selected from the ordered open connections starting from the connection with the maximum value to fulfillment of the condition when the sum of the resources of the selected connections and unoccupied resources will be sufficient to service the new connection, p After which these connections are disconnected, and the freed resource is provided to the new connection.

Images