Evaluation of Heuristic Algorithms for Optimizing Resource Assignments in LTE-Advanced
Konferenz: Mobilkommunikation – Technologien und Anwendungen - Vorträge der 19. ITG-Fachtagung
21.05.2014 - 22.05.2014 in Osnabrück, Deutschland
Tagungsband: Mobilkommunikation – Technologien und Anwendungen
Seiten: 6Sprache: EnglischTyp: PDFPersönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt
Weigelt, Bastian; Mühleisen, Maciej; Elsner, Raphael; Timm-Giel, Andreas (Institute of Communication Networks (ComNets), Hamburg University of Technology, Hamburg, Germany)
With the increasing usage of Internet data on mobile devices the infrastructure of cellular communication systems needs to be evolved constantly. LTE-Advanced (LTE-A), the latest generation of mobile communication, includes multiple degrees of freedom to achieve high data rates at low latencies. This paper evaluates methods to improve uplink resource assignment with regard to spectral efficiency. Due to a frequency reuse factor of one in LTE-A, interference between users of neighboring cells is heavily degrading system performance. With LTE-A and its Coordinated Multipoint (CoMP) techniques it is possible for base stations to exchange information and manage the allocation of frequency resources to different User Terminals (UTs) in order to mitigate interference. Deciding which UTs of different cells should transmit on the same resources leads to a Multidimensional Assignment Problem (MAP). Optimally solving the MAP (for more than two cells) is NP-hard. Therefore this paper evaluates the performance of a group of metaheuristics known as Greedy Randomized Adaptive Local Search Procedure (GRASP). These are an extension to the well-known Greedy heuristic algorithm and were introduced in 1989 by T. Feo and M. Resende. The gain of this inter-cell coordination approach is evaluated with regard to Cell Spectral Efficiency (CSE) for the overall system performance and the Cell Edge User Spectral Efficiency (CEUSE) as an indicator for fairness. Results show that GRASPs significantly improve both performance indicators. This was expected for CSE, which is the target of the optimization, but is remarkable for the CEUSE, which was not subject of the optimization.