Applying Differential Evolution to a Realistic Location Area Problem Using SUMATRA

Abstract

The mobile networks are every day more common and useful, but the costs involved in its management are high. The Location Area partitioning is a strategy of location management that tries to minimize the involved costs. In this paper we present a new approach based on Differential Evolution algorithm applied to the Location Area partitioning, as a cost optimization problem. We use realistic data [1] for generating the test networks. This work has the objective of defining the best values to the Differential Evolution (DE) parameters and setting the DE scheme, which allows us to obtain better results when compared to classical strategies and the other authors' results. The best solution is obtained after the development of four distinct experiments, each one applied to one of the differential evolution parameters. The final results obtained by this approach are very good.

Optimizing the mobility management task in networks of four world capital cities

“…As said in Section 1, these networks are based on real mobile networks and combine the goodnesses of the test networks proposed and/or studied in previous works (Subrata and Zomaya, 2003;SUMATRA;Taheri and Zomaya, 2004, 2005a,b, 2007Almeida-Luz et al, 2008, 2011, 2009aBerrocal-Plaza et al, 2013, 2012a, also solving the weaknesses of these previous problem instances. For it, we have imported the network topologies obtained in Cowling (2004) into the network simulator proposed in Taheri and Zomaya (2009).…”

“…These mobile activity traces avoid the weaknesses of the test networks presented and/or studied in previous works (Subrata and Zomaya, 2003;SUMATRA;Taheri and Zomaya, 2004, 2005a,b, 2007AlmeidaLuz et al, 2008AlmeidaLuz et al, , 2011AlmeidaLuz et al, , 2009aBerrocal-Plaza et al, 2013, 2012a, and combine their goodnesses: high number of heterogeneous network cells (Subrata and Zomaya, 2003;SUMATRA;Almeida-Luz et al, 2008, 2011, 2009aBerrocal-Plaza et al, 2013), and user's call and mobility patterns close to those that we can find in current mobile networks (Taheri and Zomaya, 2004, 2005a,b, 2007AlmeidaLuz et al, 2011AlmeidaLuz et al, , 2009bBerrocal-Plaza et al, 2012a,b). For it, we have blended the works published in Cowling (2004) and Taheri and Zomaya (2009).…”

“…there is no location update). The purpose of this study is to ensure that the traffic intensity per hour of our traces follows a bi-modal curve, as studied in Subrata and Zomaya (2003), Almeida-Luz et al (2008, 2009a, Berrocal-Plaza et al (2013), Reades et al (2007), and Louail et al (2014). Figure 6 shows the traffic intensity of these two location update strategies.…”

“…Furthermore, they implemented the following optimization algorithms: Hopfield Neural Network (HNN), Simulated Annealing (SA), Genetic Algorithm (GA), and combinations of GA with HNN. Subsequently, Almeida-Luz et al proposed the use of Differential Evolution (DE) (Almeida-Luz et al, 2008, 2011 and Scatter Search (SS) (Almeida-Luz et al, 2009a,b) to optimize the location areas planning in the SUMATRA network and in the test networks presented in Taheri and Zomaya (2004, 2005a,b, 2007.…”

Optimal Reporting Cell Planning with Binary Differential Evolution Algorithm for Location Management Problem

“…The application of Differential Evolution algorithm for cost optimization in location area, reporting cell and realistic networks is proposed in [12][13][14][15]. The best parametric values for Differential Evolution algorithm for mobile location management are studied [16,17] which are taken into consideration in the present work.…”