Auto-tuning and Self-optimization of 3G and Beyond 3G Mobile Networks

摘要: With the wireless mobile communication boom, auto-tuning and self-optimization of network parameters are more than ever key issues to provide high-quality services for end-user decrease operational expenditure operation. The special attention drawn radio resource management (RRM) is motivated by user need ubiquitous increasing complexity networks resulting from cooperation access technologies. Formerly, RRM has been based on some algorithms (admission control, allocation, handover ...) governed a set fixed thresholds. Today, procedures have undergone considerable change paradigm will shift towards completely automatic management. Optimal mechanisms, performed control methods such as fuzzy logic optimized reinforcement learning, could considerably improve functions with respect traditional parameters. This thesis introduces new results in 3G beyond networks. Auto-tuning tasks organized plane where different information exchange involved between nodes. using local loop, namely controller continuous interaction network. feeds parameter settings conversely returns its feedback delivering quality indicators indicating operating state. Different use-cases investigated. First, an allocation algorithm UMTS studied alternative existing static allocation. process dynamically adapts guard band that reserved users real time services. A best trade-off non achieved sense service becomes comparable two traffic classes especially high load situation. second use-case concerns soft For each cell receives inputs filtered downlink neighbouring cells. continually learns values learning utility function. Simulation reveal significant improvements terms performance. proposed balances base stations improves system capacity up 30% compared However, increases signalling messages interface well core negative effect minimized reducing reactivity controller. third case deals LTE (3GPP Long Term Evolution) mobility algorithm. carried out adapting margin involving couple cells according difference their loads. alleviates congestion handing off mobiles close border congested results, simplified interference models, show brings about important gain both call admission rate throughput. Finally, intersystem WLAN technologies proposed. coverage based, thresholds: first responsible - inverse direction. thresholds jointly performed. obtained adaptive tracking gain, illustrate importance intelligent provided this supported theoretical analysis extensive dynamic level simulations multi-cells' scenarios, including many relevant mechanisms impact access. do not exactly reflect reality So should be tested experimental (test-bed).