Portal de Periódicos da CAPES

Traditional Genetic Algorithm and Random-Weighted Genetic Algorithm with GIS to Plan Radio Network

2010; CRC Press; Volume: 22; Issue: 1 Linguagem: Inglês

1045-8077

Suliman Mohamed Ahmed Gaber, M. E. El-Sharkawi, M. Nour El-deen,

Advanced MIMO Systems Optimization

INTRODUCTION In the GSM network planning process, base stations are considered the most important element in the entire network because of their role in physically connecting to mobile stations (users' devices) through air interface (Ericson 2002, Nokia 2002, Mishra 2005, AirCom 2002); so the biggest problem in the planning process is where to locate those base stations appropriately to achieve the goals of service providers, including maximizing the coverage area and minimizing the construction and deployment cost (Raisanen 2006, Martin 2005). On the other hand, these base stations must be located following security and safety procedures that must prevent the citizens from suffering from the negative effects of these base stations (Gaber et al. 2009). In Egypt, the National Telecommunication Regulatory Agency (NTRA) (NTRA 2005) rolled out a protocol that governs the process of macro-cell construction to prevent the populations from the negative effects of the base stations. This protocol contains many restrictions on the process of locating base stations especially inside cities, such as the distance from the nearest school and hospital, the maximum permissible power, and other criteria to prevent human contact with the emitted radiations from base stations (Gaber et al. 2009). Many approaches are proposed to solve the problem of base stations siting; these approaches can be classified into three categories: geometric, spatial decision support systems (SDSS), and mathematical (Gaber et al. 2009). In the geometric class, the researchers tried to build a geometric model that locates base stations to minimize the negative health effects (Zhang and Fan 2004) or to solve the problem of maximizing coverage/service in a certain area (Das et al. 2006, Roy et al. 2007). Cell planning using SDSS is another direction (Raisanen 2006, Martin 2005, Scheibe 2006). Finally, in the mathematical category, cell planning is considered an optimization problem and solved using a mathematical model such as the polynomial--time approximation scheme (PTAS) (Glaber et al. 2005). Unlike the previous approaches, this paper incorporates all the factors that influence the base station siting, including cost, coverage area, and restrictions of public health. From the viewpoint of operation research, the problem of locating base stations is considered a constrained multiobjective optimization problem and can be solved using one of the famous evolutionary algorithms such as the genetic algorithms. In this paper, the base station locating problem will be solved in two ways using evolutionary algorithms. In the first algorithm, the objective functions will be aggregated to build a single combined objective function and then a traditional genetic algorithm will be incorporated with penalty functions as constraints handling (Schnier 2002, Montes and Coello 2006) to find the solution with a better fitness function value. On the other side, a random-weighted genetic algorithm will be used to find the nondominant individuals and the binary tournament selection method will be used to handle the constraints (Konak et al. 2006, Salcedo-Sanz et al. 2008). A fully functional GIS is an integration of several components and different subsystems to collect, store, retrieve, and analyze spatially referenced data. GIS is a powerful tool of acquisition, management, and analysis of spatially referenced data, but GIS is a limited tool in a spatial decision-aid domain, because, essentially, it lacks more powerful analytical tools that enable it to deal with spatial problems involving several parties with conflicting objectives and criteria (Chakhar 2003). To avoid GIS criticisms and to execute the spatial analysis on GIS data, integration between GIS and Operations Research/ Management Science (OR/MS) tools is suggested. Practically, the idea of integrating GIS with several decision support systems (DSS) tools seems to be a long-term solution. …