Telecommunications: Optimization of the Broadcasting Process in MANETs

Abstract

In this chapter, we study the application of a multi-objective cGA for optimizing DFCN, a broadcasting protocol specially designed for metropolitan ad hoc networks (MANETs). Optimizing DFCN is a newly defined problem {[}18] consisting of finding the best values for a set of important parameters of the protocol that characterize its behavior. The objectives are to minimize the use of the network and the total process time, and to maximize on the other hand the coverage of the broadcasting protocol (the number of devices reached). Mobile Ad Hoc Networks (MANETs) are fluctuating networks populated by a set of communicating devices (also called nodes) which can spontaneously interconnect each other without any pre-existing infrastructure. This means that no organization is present in such networks. The most popular wireless networking technologies available nowadays for building MANETs are Bluetooth and IEEE802.11 (WiFi). This implies that a) devices communicate within a limited range, and b) devices may move while communicating. A consequence of mobility is that the topology of such networks may change quickly and in unpredictable ways. This dynamical behavior constitutes one of the main obstacles for performing efficient communications. In this chapter we are considering the problem of broadcasting on a particular subclass of MANETs called Metropolitan MANETs, which have some specific properties: the density in the network is heterogeneous and dynamic (particularly, high density regions do not remain active full time). The broadcasting strategy we are considering in this work is the so called Delayed Flooding with Cumulative Neighborhood protocol (DCFN) {[}132]. Three real world examples of such networks, a shopping mall, a metropolitan area, and a highway environment, have been taken into account so that, instead of providing a multi-purpose protocol, the originality of our proposal lies in tuning the broadcasting service for each particular network. Optimizing a broadcasting strategy implies multiple goals to be satisfied at the same time, such as maximizing the number of devices reached (coverage), minimizing the network use (bandwidth), and/or minimizing the duration of the process. Thus, what we are facing is known as a multi-objective optimization problem {[}65, 49], and we propose a new multi-objective cGA (cMOGA) for solving it. The results will be compared versus the main state-of-the-art algorithm in multi-objective optimization, NSGA-II {[}67]. As it was already mentioned in Chap. 9, the most popular algorithms for solving multi-objective problems are evolutionary algorithms {[}49, 65]. Though, very few works use GAs based on cellular models {[}146, 159, 190], even though cGAs have demonstrated to have very high efficiency and accuracy in mono-objective optimization {[}12, 17, 19, 99]. The proposed algorithm, cMOGA, represents a contribution to this field. Furthermore, this research line is, at the very best of our knowledge, the first attempt to solve the broadcasting problem on MANETs using a multi-objective EA. The rest of this chapter is organized in the following manner. In Sect. 14.1 we describe the considered problem, the set of proposed scenarios for this problem, and the broadcasting strategy we plan to optimize. The proposed approach (a multi-objective cGA) is described in Sect 14.2. In Sect. 14.3 we present our experimentation and analyze the results. Our proposed algorithm is compared versus NSGA-II ( the main state-of-the-art algorithm in the multi-objective domain) in Sect. 14.4. We finish this chapter with our main conclusions.