Optimization of femtocell networks using heuristic search techniques

Abstract

In telecommunications, a femtocell is a low power small base station deployed by end users. This technology is expected to improve both coverage and capacity, especially indoors. Providers of cellular services are increasingly implementing this technology in their networks. However, the decentralized manner of femtocells operation introduces several challenges. This thesis presents an overview of femtocells, highlights the network requirements for femtocell deployments, and discusses the major issues that affect its operation and performance. It also describes the key challenges of deploying femtocells and provides some of the proposed solutions. In addition, deploying a number of femtocells to jointly provide coverage in an enterprise environment raises critical challenges especially in future self-organizing networks which rely on plug-and-play fashions. This research is focused on an optimization technique based on the genetic algorithm (GA) for a centralized self-optimizing network containing a group of femtocells. In order to optimize the network coverage in terms of load handling, coverage gaps, and overlaps. The algorithm provides a dynamic update of the downlink pilot powers of the deployed femtocells. The results show the ability of the algorithm to optimize the coverage effectively according to the global traffic distribution and the levels of interference among neighboring femtocells. The proposed algorithm is also compared with the fixed pilot power scheme. The results show an appreciable reduction in pilot power pollution and a significant enhancement in network performance. Moreover, the solution quality and the efficiency of the algorithm are evaluated by comparing the results generated by an exhaustive search with the pilot power configuration obtained from the GA. The research also evaluates the performance of genetic algorithm with two other heuristic techniques, particle swarm optimization (PSO) and simulated annealing (SA), when employed to solve the same problem. The differences in the behaviors of the proposed algorithms are presented. The results show that genetic algorithm and particle swarm have a higher potential to solve the problem compared to simulated annealing. This is due to their faster convergence time which is an important parameter for dynamic update of femtocells. Finally, the research investigates the performance of a clustering-based femtocell coverage optimization. A cluster head is chosen depending on a specific criterion to perform the optimization for the femtocells attached to it. The network performance of this clustering approach was also assessed.

Date of Award	2012
Original language	American English
Supervisor	Mahmoud Al Qutayri (Supervisor)