Optimization of Uplink Sum-Rate for Bin Based Clustered Cellular System Using a Genetic Algorithm

Majid, M. I., Imran, M. A. and Hoshyar, R. (2010) Optimization of Uplink Sum-Rate for Bin Based Clustered Cellular System Using a Genetic Algorithm. In: Proceedings of the 6th International Wireless Communications and Mobile Computing Conference (IWCMC '10), Caen, France, 28 Jun - 02 Jul 2010, pp. 1016-1020. ISBN 9781450300629 (doi:10.1145/1815396.1815628)

Full text not currently available from Enlighten.


Efficient use of available spectrum is of concern to future wireless network planners. Although global cooperation at access points (APs) maximizes sum rate, for large networks this assumption is too complex to implement. We partition the wireless network into localized jointly decoded cells implemented as fixed size clusters. This is an alternative model which is more practical and improves efficiency of current systems. Conventionally, frequency allocation using interference avoidance maximizes spectrum usage. However, with clusters, careful allocation of interference needs to be explored. This is done using flexible bin based frequency allocation and applying heuristic tools. This work is the first known attempt to analyze uplink capacity of bin based fixed cluster cellular systems using genetic algorithms. To implement this, we derive an expression for the uplink capacity of bin based fixed clusters. We then input this as a fitness function to a modified simple genetic algorithm to compute a good fit for our bin allocation problem. We deduce that for sparsely distributed APs and large cluster sizes, rates close to that of a joint processor are achievable. Moreover, decreasing AP density for small cluster sizes (inter cell distance greater than 5 km for 7 cell-cluster) has insignificant effect on sum rate performance. However, with a nominal increase in the number of bins available for transmission, for dense system, the per-cell sum rate of a clustered cellular system can reach close to that of a hyper receiver using a genetic algorithm.

Item Type:Conference Proceedings
Additional Information:The first author is a Commonwealth PhD scholar and was supported by CSFP, UK.
Glasgow Author(s) Enlighten ID:Imran, Professor Muhammad
Authors: Majid, M. I., Imran, M. A., and Hoshyar, R.
College/School:College of Science and Engineering > School of Engineering

University Staff: Request a correction | Enlighten Editors: Update this record