Mathematical optimisation and signal processing techniques in wireless relay networks
thesisposted on 2012-09-07, 12:04 authored by Ranaji Krishna
With the growth of wireless networks such as sensor networks and mesh networks, the challenges of sustaining higher data rates and coverage, coupled with requirement for high quality of services, need to be addressed. The use of spatial diversity proves to be an attractive option due to its ability to significantly enhance network performance without additional bandwidth or transmission power. This thesis proposes the use of cooperative wireless relays to improvise spatial diversity in wireless sensor networks and wireless mesh networks. Cooperation in this context implies that the signals are exchanged between relays for optimal performance. The network gains realised using the proposed cooperative relays for signal forwarding are significantly large, advocating the utilisation of cooperation amongst relays. The work begins with proposing a minimum mean square error (MMSE) based relaying strategy that provides improvement in bit error rate. A simplified algorithm has been developed to calculate the roots of a polynomial equation. Following this work, a novel signal forwarding technique based on convex optimisation techniques is proposed which attains specific quality of services for end users with minimal transmission power at the relays. Quantisation of signals passed between relays has been considered in the optimisation framework. Finally, a reduced complexity scheme together with a more realistic algorithm incorporating per relay node power constraints is proposed. This optimisation framework is extended to a cognitive radio environment where relays in a secondary network forward signals without causing harmful interferences to primary network users.
- Mechanical, Electrical and Manufacturing Engineering
Publisher© Ranaji Krishna
NotesA Doctoral Thesis. Submitted in partial fulfillment of the requirements for the award of Doctor of Philosophy of Loughborough University.
EThOS Persistent IDuk.bl.ethos.519722