RAT selection algorithims for common radio resource management

Publication Type:
Thesis
Issue Date:
2011
Full metadata record
Files in This Item:
Filename Description Size
Thumbnail01front.pdf418.7 kB
Adobe PDF
Thumbnail02whole.pdf3.62 MB
Adobe PDF
The future wireless network is expected to be a heterogeneous network, which integrates different Radio Access Technologies (RATs) through a common platform. A major challenge arising from the heterogeneous network is Radio Resource Management (RRM) strategy. Common RRM (CRRM) has been proposed in the literature to jointly manage radio resources among a number of overlapped RATs in an optimized way. RAT selection algorithm is one of the key research areas in CRRM. In the literature, a number of RAT selection algorithms have been proposed and some performance evaluations have been conducted. However, this area still has many challenges. Some performance metrics still have not been evaluated well and the existing algorithms can be further improved. In this thesis, some performance evaluations on a number of RAT selection algorithms have been carried out. The effects of load threshold setting on Load Balancing (LB) based RAT selection algorithm’s performance are evaluated. It is found that setting a proper load threshold can achieve a more balanced load distribution among overlapped cells. However, it will also cause higher Direct Retry (DR)/Vertical Handover (VHO) probability and in turn higher overhead and blocking/dropping probability. This thesis evaluates the performance of three RAT selection algorithms, LB based using maximum resource consumption, LB based using minimum resource consumption, and service based algorithms, in terms of traffic distribution, blocking probability, throughput, and throughput fairness for a co-located GERAN/UTRAN/WLAN network. Simulation results show that in terms of blocking probability, the service based algorithm is the worst one when the traffic load is high. In terms of data throughput, the LB based using maximum resource consumption algorithm performs better than the other two when the traffic load is low. However, the service based algorithm outperforms the other two when the traffic load is high. In terms of throughput fairness, the service based algorithm achieves the best performance. The relationship among overall downlink data throughput, user satisfaction rate, and path loss threshold is studied in this thesis. It is found that in some cases, an optimum path loss threshold value can be found to achieve better performance in terms of both overall throughput and user satisfaction rate. However, in other cases, a tradeoff has to be made between them. This thesis studies policy based RAT selection algorithms for a co-located UMTS/GSM network. A three-complex policy based algorithm called IN*VG*Load algorithm is proposed based on improvements on the existing IN*VG algorithm. The simulation results show that the IN*VG*Load algorithm can optimize the system performance in highly loaded co-located UMTS/GSM networks. A Proposed Policy Based Algorithm 2 is found to be suitable for low to medium loaded UMTS/GSM networks. In order to support the conceptual development of RAT selection algorithms in heterogeneous networks, the theory of Markov model is used. This thesis proposes both user level and network level Markov models for a co-located GERAN/UTRAN/ WLAN network. The proposed Markov models are not only extensions of the existing two co-located RATs models but more complex with more state transitions. The performance of two basic RAT selection algorithms: LB based and service based algorithms are evaluated in terms of call blocking probability. The numerical results obtained from the proposed network level Markov model are validated by simulation results.
Please use this identifier to cite or link to this item: