Energy efficiency and interference management in long term evolution-advanced networks.

Abstract

Cellular networks are continuously undergoing fast extraordinary evolution to overcome technological challenges. The fourth generation (4G) or Long Term Evolution-Advanced (LTE-Advanced) networks offer improvements in performance through increase in network density, while allowing self-organisation and self-healing. The LTE-Advanced architecture is heterogeneous, consisting of different radio access technologies (RATs), such as macrocell, smallcells, cooperative relay nodes (RNs), having various capabilities, and coexisting in the same geographical coverage area. These network improvements come with different challenges that affect users’ quality of service (QoS) and network performance. These challenges include; interference management, high energy consumption and poor coverage of marginal users. Hence, developing mitigation schemes for these identified challenges is the focus of this thesis. The exponential growth of mobile broadband data usage and poor networks’ performance along the cell edges, result in a large increase of the energy consumption for both base stations (BSs) and users. This due to improper RN placement or deployment that creates severe inter-cell and intracell interferences in the networks. It is therefore, necessary to investigate appropriate RN placement techniques which offer efficient coverage extension while reducing energy consumption and mitigating interference in LTE-Advanced femtocell networks. This work proposes energy efficient and optimal RN placement (EEORNP) algorithm based on greedy algorithm to assure improved and effective coverage extension. The performance of the proposed algorithm is investigated in terms of coverage percentage and number of RN needed to cover marginalised users and found to outperform other RN placement schemes. Transceiver design has gained importance as one of the effective tools of interference management. Centralised transceiver design techniques have been used to improve network performance for LTE-Advanced networks in terms of mean square error (MSE), bit error rate (BER) and sum-rate. The centralised transceiver design techniques are not effective and computationally feasible for distributed cooperative heterogeneous networks, the systems considered in this thesis. This work proposes decentralised transceivers design based on the least-square (LS) and minimum MSE (MMSE) pilot-aided channel estimations for interference management in uplink LTE-Advanced femtocell networks. The decentralised transceiver algorithms are designed for the femtocells, the macrocell user equipments (MUEs), RNs and the cell edge macrocell UEs (CUEs) in the half-duplex cooperative relaying systems. The BER performances of the proposed algorithms with the effect of channel estimation are investigated. Finally, the EE optimisation is investigated in half-duplex multi-user multiple-input multiple-output (MU-MIMO) relay systems. The EE optimisation is divided into sub-optimal EE problems due to the distributed architecture of the MU-MIMO relay systems. The decentralised approach is employed to design the transceivers such as MUEs, CUEs, RN and femtocells for the different sub-optimal EE problems. The EE objective functions are formulated as convex optimisation problems subject to the QoS and transmit powers constraints in case of perfect channel state information (CSI). The non-convexity of the formulated EE optimisation problems is surmounted by introducing the EE parameter substractive function into each proposed algorithms. These EE parameters are updated using the Dinkelbach’s algorithm. The EE optimisation of the proposed algorithms is achieved after finding the optimal transceivers where the unknown interference terms in the transmit signals are designed with the zero-forcing (ZF) assumption and estimation errors are added to improve the EE performances. With the aid of simulation results, the performance of the proposed decentralised schemes are derived in terms of average EE evaluation and found to be better than existing algorithms.