An adaptive congestion control mechanism for intelligent broadband networks
Introduction
In a typical intelligent network-based configuration (IN), the key elements that participate in the establishment and release of an IN call are the service switching point (SSP), the service control point (SCP) and the intelligent peripheral (IP). In the most simple case of a broadband service switching point network (termed B-SSP in the rest of this paper) can be an ATM switch which is enhanced with very basic IN capabilities, that is to discriminate IN from no-IN calls and to forward them to the broadband service control point (B-SCP) for further processing [1]. The broadband intelligent peripheral (B-IP) in case it is not an adjunct to the B-SSP will be responsible for the interaction between the B-SCP and the user. The B-SCP is the main actor in the provision of complex IN services. It contains the data and service logic required for the processing of IN calls and acts as the initiator of the actions required to establish bearer connections for accommodating the communication needs of the service. This centralized approach which is reflected in the multiplicity of tasks undertaken by the B-SCP in a broadband IN architecture, makes the B-SCP design a crucial engineering task.
The above remark becomes more intense in the light of the novel multimedia nature of services like interactive multimedia retrieval (IMR), and Broadband Video Conference, which call for even more complex call processing procedures at the B-SCP [2]. For these services the frequency of interactions between the B-SSP and B-SCP extends to higher values, a situation which appears as a dramatic increase of the already heavy signalling traffic of the network. Hence, it is natural to expect, that in future broadband multimedia networks exploiting the IN technology, the B-SCP will have to face not only the huge amount of rapidly increasing requests for multimedia services, but also, the heavy processing load brought about in each of the calls. Things become even more critical if we consider the usual situation of IN-based ISDN configurations where a single SCP serves a large number of SSPs [3]. Therefore, when the B-SCP load becomes higher than its capacity, performance degrades and eventually user requests experience long delays or in the worst case a high percentage of calls are rejected.
From the discussion above it became obvious that when a centralized approach is taken in the design of an IN system the presence of a congestion control mechanism is inevitable for handling overload conditions and allowing the normal operation of the network under heavy load. Congestion control becomes even more important with reference to the scalability an IN system must present in terms of an increase in the number of connected users. The necessity for scalable systems has pushed research on IN technology further to the adoption of Mobile Agent Technologies and Distributed Processing Environment that allow the distribution of intelligence from the B-SCP to the B-SSPs, on a need to have basis [18].
Coming back to the congestion control issue, recent research has shown that message discard alone is a poor overload mechanism, and therefore some form of feedback is desirable [5]. If feedback control is not used, the call completion rate may fall to zero for reasonable retry probabilities. In this account, a feedback congestion control mechanism is used by MTP3 layer of SS7 [6]. Transfer control congestion messages are sent back to the source reducing the offered load to the affected destination. However, although congestion control mechanisms have been widely investigated for SS7-based networks [6], [7], [8], analogous issues have not been stressed extensively in the context of intelligent broadband networks. Even though the global distributed plane (GDP) of the IN conceptual model provides congestion control capabilities, the mechanisms for accomplishing these tasks are not yet defined.
In this paper, we investigate ways to protect an intelligent broadband network from overload problems experienced in the B-SCP and propose an adaptive congestion control mechanism (ACCM). ACCM differs from other proposals covering the ISDN case (see [9]) in that it does not use tabulated static information for determining the gap duration corresponding to the transmission of a single message or call but adapts the duration and the number of ON and OFF periods within the gap dynamically based on the current traffic load. A result of this is that the adaptation of the ON and OFF periods takes place immediately without any additional delay. Furthermore, our evaluation study is not restricted to the use of a single source destination pair but simulates the entire network. To do so, models that represent complex higher level broadband signalling protocols and IN functional entities are developed. As it has been earlier mentioned the B-SCP is expected to be the main point of overload conditions so the proposed mechanism is primarily applied between the B-SCP and the B-SSP. Results obtained by simulation show that apart from the B-SCP, congestion may occasionally strike other network elements like the B-SSPs, so the ACCM is further extended to the B-SSP side.
The paper is organized as follows. In Section 2, we present the functional entities and protocols of the intelligent broadband network as this has been specified in the context of a three-year transnational research project [19] and has been documented in [1]. A detailed description of the ACCM is given in Section 3 and flowchart diagrams in Appendix A. Section 4 together with Appendix B contains the set of performance models used to describe the network entities and the exchange of control (either signalling or IN messages) among them. Section 5 includes a detailed study on the effect of the variation of ACCM parameters on network performance. The proposed mechanism is evaluated via simulation in Section 6 and results are recapitulated in Section 7.
Section snippets
Physical/functional entities and protocols of the intelligent broadband network
The intelligent broadband network is an extension of the typical broadband signalling network configuration to include intelligent entities providing the additional capabilities required for the processing of IN calls. The IN physical entities that participate in the establishment and release phase of a call are the B-SSP, the B-SCP and the B-IP. The communication path includes also signalling transfer points (STP), which are typical broadband switches with signalling capabilities and
Adaptive congestion control mechanism
The ACCM considered in this paper is based on the call gap capabilities offered by GDP [13]. The call gapping is network initiated. The proposed mechanism could be applied between nodes of the same type like B-SSPs or different types like a B-SSP and a B-SCP. In our study, the mechanism is applied between the B-SSP and the B-SCP and between the STP1 and the B-SSP. For sake of generality the mechanism is described as executed between node-S (source node) and node-D (destination node). We assume
Models
In this section, we present the set of models developed to evaluate the system performance under the proposed congestion mechanism. For each of the physical entities involved in a call set-up/release request of an IN service, a model is developed. The model includes the protocol layers and functional entities, as well as the flow of information among them. We assume that a single processor serves the entire node and that time sharing among protocols is exercised by a FIFO scheme. Since the
Derivation of ACCM parameters
To investigate how the values of the mechanism parameters affect the performance of the network and furthermore to select appropriate values for the ACCM parameters, we will study the system behaviour under overload conditions. The point at which the system becomes overloaded is defined by calculating the theoretical maximum throughput of the network. We then consider loads exceeding this maximum throughput and vary the parameter values.
Experimental evaluation of the effectiveness of the ACCM
This section presents five scenarios in which we progressively apply the proposed mechanism to the network nodes in order to reveal the positive effect of the mechanism in network performance, during overload conditions. Several network architectures have been examined and extensive simulations have been performed. The values of the congestion parameters are taken from Section 5 and are shown in Table 6. All times are in ms and are comparable to the process times of functional entities and
Conclusions
We have proposed an ACCM, which can be applied at different nodes of an IN-based broadband network and protect the network elements from overload conditions. The proposed mechanism is activated in order to prevent overload conditions at the candidates for overload nodes. When overload conditions are inevitable due to the extremely high offered load the proposed mechanism protects the overload node by reducing the rate of message transmission. In an IN-based broadband network the ACCM can be
Acknowledgements
This work was partially funded by the EU ACTS Project INSIGNIA (AC-068) [19]. The opinions appearing in this paper are those of the authors and not necessarily of the other members of the project consortium.
Spyros E. Polykalas was born in Kefalonia, Greece, in 1971. He received the Dipl.-Ing. degree from the Electrical Engineering Department of the University of Patras, Greece in 1994 and the Ph.D. degree from the Electrical Department of the National Technical University of Athens in 1999. His research interests are in the area of signalling systems, Intelligent Networks, Active Networks, Mobile communications and performance evaluation. He has participated in several European Union ACTS and
References (21)
- et al.
Parallel architectures for processing high speed network signalling protocols
IEEE/ACM Trans. Netw.
(1995) Ensuring robust call throughput and fairness for SCP overload controls
IEEE/ACM Trans. Netw.
(1995)Effects of feedback delay on the performance of the transfer-controlled procedure in controlling CCS network overloads
IEEE J. Select. Areas Commun.
(1994)On the efficiency of using the transfer-controlled procedure during periods of STP processor overload in SS7 networks
IEEE J. Select. Areas Commun.
(1994)- et al.
Performance analysis of SS7 congestion controls under sustained overload
IEEE J. Select. Areas Commun.
(1994) - et al.
A scalable and reusable emulator for evaluating the performance of SS7 networks
IEEE J. Select. Areas Commun.
(1994) - et al.
A control mechanism to prevent orrelated message arrivals from degrading signaling No. 7 network performance
IEEE J. Select. Areas Commun.
(1994) - et al.
Service control point overload rules to protect intelligent network services
IEEE/ACM Trans. Netw.
(1998) - et al.
Capacity and performance analysis of signaling networks in multivendor environments
IEEE J. Select. Areas Commun.
(1994)
Cited by (7)
Overload control for the wireless intelligent network
2014, Journal of NetworksApproach to congestion pattern analysis based on Bayesian Network
2010, 2010 International Conference on E-Health Networking, Digital Ecosystems and Technologies, EDT 2010Intelligent service influence evaluation for SIP proxy server performance
2006, Proceedings of the International Conference on Information Technology Interfaces, ITIA middleware service for employing and evaluating resource management algorithms in distributed networks
2005, International Journal of Communication Systems
Spyros E. Polykalas was born in Kefalonia, Greece, in 1971. He received the Dipl.-Ing. degree from the Electrical Engineering Department of the University of Patras, Greece in 1994 and the Ph.D. degree from the Electrical Department of the National Technical University of Athens in 1999. His research interests are in the area of signalling systems, Intelligent Networks, Active Networks, Mobile communications and performance evaluation. He has participated in several European Union ACTS and National Research projects. He has received several national awards for the entire five years of his undergraduate study. Dr. Polykalas is a member of IEEE and the Technical Chamber of Greece.
George T. Kolyvas was born in Corfu, Greece, on 5th August 1970. He received the Dipl.-Ing. degree from the Electrical Engineering Department of the University of Patras, Patras, Greece in 1993 and the Ph.D. degree from the Electrical and Computer Engineering Department of the National Technical University of Athens (NTUA), Athens, Greece in 1997. Dr. Kolyvas has been involved in several European Community and National Research Projects and his research interests are in the area of B-ISDN access networks, signalling systems, multimedia service design, Intelligent Network technology, Information Systems, telematics applications, performance evaluation and queuing theory. Dr. Kolyvas has received several national awards for the entire five years of his undergraduate study. He is a member of IEEE (’96) and the Technical Chamber of Greece (’93).
Iakovos S. Venieris was born in Naxos, Greece, on 3rd March 1965. He received the Dipl.-Ing. degree from the University of Patras, Patras, Greece in 1988, and the Ph.D. degree from the National Technical University of Athens (NTUA), Athens, Greece, in 1990, all in electrical & computer engineering. He is currently an Associate Professor in the Electrical and Computer Engineering Department of NTUA. His research interests are in the fields of B-ISDN, Intelligent Networks, Internet, distributed processing, service and network control, and performance evaluation. He has several publications in the above areas. Dr. Venieris has received several national and international awards for academic achievement. He has been exposed to standardisation body work and has contributed to ETSI and ITU-T. He has participated in several European Union and national projects. He is member of IEEE and the Technical Chamber of Greece.