OPTIMAL AND PRACTICAL HANDOVER DECISION ALGORITHMS IN HETEROGENEOUS MACRO-FEMTO CELLULAR NETWORKS
İsmail Cem Atalay
Electronics Engineering, MSc Program, 2012
Assoc. Prof. Özgür Gürbüz (Thesis Supervisor), Assoc. Prof. Özgür Erçetin, Assoc. Prof. Albert Levi, Asst. Prof. Kerem Bülbül, Assoc. Prof. Erkay Savaş
Date & Time: August 7th, 2012 – 09:30
Place: FENS G032
Driven by the smart tablet/phone revolution and the proliferation of bandwidth hungry applications such as cloud computing and streaming video, the demand for high data rate wireless communication is increasing tremendously. In order to meet the increasing demand from subscribers, wireless operators are in the process of augmenting their macrocell network with supplemental infrastructure such as microcells, distributed antennas and relays. An alternative with lower upfront costs is to improve indoor coverage and capacity by using end-consumer installed femtocells. A femtocell is a low power, short range (up to 100 meters coverage radius) cellular wireless access point (AP), functioning in service provider owned licensed spectrum. Due to the proximity of end users to the femtocell access points, APs are able to provide higher end-user QoE (quality of experience) and better spatial reuse of limited spectrum. Femtocells are useful in offloading the macro-cellular network as well as reducing the operating and capital expenditure costs for operators.
Femtocells coexist with legacy cellular networks consisting of macrocells. In this emerging combined architecture, large number of Femtocell Application Point (FAPs) is randomly deployed in the coverage area of macro BSs. However, several problems related to MM (mobility management) and RM (resource management) in this combined architecture still remain to be solved. The ad hoc deployment of FAPs and asymmetric radio communication and call processing capabilities between macro-femto networks are the primary causes of these problems. Uncoordinated deployment of FAPs providing indoor oriented wireless access service within the macro coverage may cause severe interference problems that need to be mitigated and handled by RM/MM schemes. The MM decisions should take into account the resource constraints and UE (user equipment) mobility in order to prevent unnecessary or undesirable handovers towards femtocells. Ignoring these factors in MM decisions may lead to low customer satisfaction due to mismanagement of handover events in the combined macro-femto network, delayed signaling traffic and unsatisfactory call/connection quality.
In order to address all of the aforementioned issues, the handover decision problem in combined femto-macro networks has been formulated as a multi-objective non-linear optimization problem. Since there are no known analytical solution to this problem, an MDP (Markov Decision Process) based heuristic has been proposed as a practical and optimal HO decision making scheme. This heuristic has been updated and improved in an iterative manner and has also been supported by a dynamic SON (Self Organizing Networks) algorithms that is based on heuristic's components. The performance results show that the final version of MDP based heuristic has significantly superior performance in terms offloading the macro network, minimizing the undesirable network events (e.g. outage and admission rejection) when compared to state-of-art handover algorithms.