Decision Choice Optimization With Genetic Algorithm in Communication Networks

Introduction

In the recent years, the privatization and liberation of services in telecommunication networks sectors lead to a diversification of services providers. This diversity leads to strong competition between them, each of which tries to attract and / or retain customers. Since do not give true information about their systems (client confusion), customers do not have complete information to make a good decision. This confusion presents an obstacle to customers to have all the information on the operator’s offer. For it, consumer confusion has been studied as an important in many markets due to dynamics, such as over choice, excessive marketing communications, and similar tariffs and promotions. Hence, in the presence of confusion, the customer’s choice is often uncertain (bounded rationality).

Telecommunication networks also have been considered in the economic literature as one of the sectors that causes confusion among consumers (Ait Omar, Outanoute, Baslam, Fakir, & Bouikhalne, 2017) Complex fee structures introduced by service providers, number portability, loyalty programs and complicated service variations make it difficult for consumers to understand the type of transaction and get the best report price /QoS.

The problems related to the choice of an operator is on several parameters including real quality of service, theoretical quality of service, bandwidth, price... Operators decide a price and QoS for services offered to its customers. QoS proposed remains a parameter that depends on other variables namely bandwidth, the share of this market operator. In these circumstances, an operator can never guarantee the quality of service it promises to customers. We call it then theoretical QoS, while the QoS perceived by customers is the real QoS. In the telecommunications networks, the credibility of each operator is measured by the difference between its theoretical QoS and the real QoS. A customer is interested in the recognition of with good credibility (which has a real QoS close to the theoretical QoS). In addition, as there is no real QoS to operators cannot solve this problem in a direct way. Hence, this kind of problem can be modeled within the meaning of the inverse problem. The inverse problem is generally ill-posed problem, on the contrary live with the problem said that the solution exists, is unique and depends on data. For example, if it is to reconstruct the past state of a system knowing its current state, we are dealing with an inverse problem; but the fact of predicting the future state given the current state is a direct problem. Similarly, in the case of a determination of parameters of a system knowing a part of the stage (a part of the set of parameters); we speak of parameter identification problems.

In related work, many studies have contributed to the study of the problem of consumer confusion and their bounded rationality. However, to my knowledge there is no work that is interested in finding the form of real QoS to know the confidence level of service providers in telecommunication networks. The authors of the paper (Ait Omar et al., 2017) they proposed a model that is based on the economic weighing of and the Luce probabilistic choice model. They concluded that the client is confused or not by the SPs, if its degree of irrationality increases, the earns more; this model of Luce’s choice is often more complicated than the discrete choice model implied by the authors of the paper ((Coucheney, Maille, & Tuffin, 2013),(Ait Omar, Garmani, El Amrani, Baslam, & Fakir, 2019)). In addition, the authors of the book (Lorkowski & Kreinovich, 2018) debate the confusion concept, display that irrational decision-making can be explained if one takes into account that human capacities to process information are limited. Thus, they used and improved heuristic techniques to predict the quality of the decision by formulating and solving optimization problems. The use of heuristic methods for the optimization of the proposed models is due to their efficiency in the search for the global optimum.

The instead of this chapter is organized as follows: in section 2, we present modeling problematic using the inverse problem theory. We present in section 3 the genetic algorithm as optimization method used to solve the models proposed in our work. In section 4, we present the different numerical results obtained. We conclude this study with perspectives in section 5.