An empirical study for an improvement of cellular network using genetic algorithm

Abstract

INTRODUCTIONThe aim of planning and management in mobile cellular networks is to design, organize, and manage the cellular networks to reduce the cost of operation and meet the quality of service required by the subscribers. The efficiency of the network is influenced by the location management of mobile cellular networks. Location management deals with how to keep track of an active mobile station. Location registration (i.e., location update) and paging are the essential factors in the location management. Partition the network area into a set of Location Areas is the common strategy applied to increase the efficiency of the network [2]. In this strategy, each location area has one or more number of cells. The goal is to partition the network into a given number of location areas such that the total paging and location update cost is maintained at its minimum. In such partitioning approach, a balance between location update cost and paging cost is the main objective in order to enhance the performance of the network [2]. In particular, upon the arrival of a mobile-terminated call, the system tries to find the Mobile Terminal (MT) by searching for it among a set of Base Transceiver Stations (BTSs) over pre-assigned region. This search is called paging, and the set of cells in which a mobile is paged is called the Location Area (LA). At each LA boundary crossing, TMs in all cells update their new locations, and this is called location update (LU). Therefore, minimizing the cost of paging and location update depends on the number of cells assigned in each LA. Due to high mobility and increase of subscribers, most of recent mobile network systems, including the GSM system, are employing the zone based scheme [2]. In this scheme, the service area is divided into groups of cells forming LAs. Finding the optimal number of cells assigned in each LA and finding number of location areas has motivated many researchers, as it is a difficult combinatorial optimization problem which is classified as an NP-hard problem [6]. Metaheuristic approaches such as Simulated Annealing algorithm [7], Particle Swarm Optimization [8], and Ant Colony System [1] have addressed the LA planning problem for the zone based scheme. However, clustering of cells based on frequencies is not considered in these approaches. Due to a limited frequency bandwidth, the cells need to be clustered in terms of the frequencies pattern such as in Global System for Mobile communication (GSM) networks [5]. In particular, clustering constraint raised by limited frequency bandwidth should be taken into consideration in the LA planning. In this paper, the Genetic Algorithm is adapted to solve the LA planning problem for an existing network. The clustering constraint will be considered in this work along with other realistic constraints. The rest of the paper is organized as follows: Section II provides problem formulation. Section III presents the adaptation of the genetic algorithm for the LA planning problem. Section IV provides a set of computational experime...