Causal and Anticipative Models for the Dimensioning of 3G Multiservice Networks
We study mathematical models and discuss optimization algorithms for the dimensioning of 3G multimedia networks. We propose two models which aims at dimensioning networks with both a radio and a core component. The ﬁrst one is an anticipative one in which we assume that we know a priori the trafﬁc over the planning period and the dimensioning is deﬁned with a best possible call admission control procedure. The second one is a causal one in which we deﬁne an explicit call admission control procedure which makes the accept/reject decisions without any knowledge on the forthcoming trafﬁc. We then compare, on an experimental basis, the dimensioning obtained by both models on some multi-service networks.
Dimensioning of 3G multi-service network offers new challenges as it involves handshake between wired and radio networks, that both obey to different rules and protocols. Although dimensioning is a well studied topic in telecommunication, few papers have yet been published in which the authors consider a network with both a radio and a wired component. Most papers are studying network dimensioning under particular aspects, e.g., with single path routing (e.g., or dynamic/adaptive routing (e.g., with single service or multi-service with single hour load or multi-hour trafﬁc (e.g., Dutta and Lim ), … The main concern of radio dimensioning is the evaluation of an uplink/downlink cell capacity. The basis for cell capacity computation can be found in for CDMA uplink, and in for WCDMA uplink/downlink in UTMS networks, and in recent papers such as, e.g., Cordier and Ortega (downlink). In the current paper, we consider dimensioning of a 3G multi-service network with single path routing and we design two multi-hour mathematical models that take into account most of the basic network operations, as well as the QoS (Quality of Service) and the GoS (Grade of Service) requirements. Both the QoS constraints and the GoS thresholds depend on the service classes and on the user priorities. For instance, we assume that each service class requires a ﬁxed peak rate allocation of bandwidth for the duration of a connection, and that the various services, depending on the user priorities, have different peak rates.