Engineering and Information Technology Collected Works - Theses
Permanent URI for this collection
Search Results
1 - 5 of 5
-
ItemBurst erasure correction convolutional codes(University of Melbourne, 2010)
-
ItemA new approach to incremental processing of nearest neighbor queries(University of Melbourne, 2009)
-
ItemGIS obfuscation system for location privacy in LBS(University of Melbourne, 2008)
-
ItemResource optimization in multiuser communication networks(University of Melbourne, 2006)The focus of this thesis is on realizable techniques, algorithms and protocols to optimize specified resources of communication networks that support multiple users. Interactions between such users, through interference effects for example, are specifically incorporated into the optimization framework. The thesis is presented in two parts. In Part I we focus on wireless networks. We develop a new framework for the joint optimization of all mobiles' transmit power and linear receivers, applied to interference limited CDMA cellular networks subject to fast-fading. Our framework involves conceptually simple algorithms that operate on the large-scale fading time-scale to ensure quality of service (QoS) constraints on outage probability are met with minimal power. Similar QoS constraints are applied to a mobile ad-hoc network in the context of a cross-layer design. We develop new convexity results surrounding the Shannon link capacity to subsequently solve the nonconvex design formulation with a novel technique involving a sequence of convex relaxations. A globally optimum protocol is then provided for the cross-layer design problem of transmitter power and source-rate allocation that preserves the existing TCP stack for rate-allocation. In Part II we focus on multicarrier systems. In particular, algorithms and protocols for dynamic spectrum management of digital subscriber line (DSL) access networks are developed. We derive a protocol called SCALE that jointly optimizes the power spectrum of all DSLs in a distributed manner having low complexity implementation. These developments give rise to new water-filling techniques devoid of bisection or channel sorting steps, that are particularly useful to applications of iterative water-filling. Lastly, novel algorithms for band preference design are developed to enhance the performance of current-day DSL networks.
-
ItemResource optimization in multiuser communication networks(University of Melbourne, 2006)The focus of this thesis is on realizable techniques, algorithms and protocols to optimize specified resources of communication networks that support multiple users. Interactions between such users, through interference effects for example, are specifically incorporated into the optimization framework. The thesis is presented in two parts. In Part I we focus on wireless networks. We develop a new framework for the joint optimization of all mobiles' transmit power and linear receivers, applied to interference limited CDMA cellular networks subject to fast-fading. Our framework involves conceptually simple algorithms that operate on the large-scale fading time-scale to ensure quality of service (QoS) constraints on outage probability are met with minimal power. Similar QoS constraints are applied to a mobile ad-hoc network in the context of a cross-layer design. We develop new convexity results surrounding the Shannon link capacity to subsequently solve the nonconvex design formulation with a novel technique involving a sequence of convex relaxations. A globally optimum protocol is then provided for the cross-layer design problem of transmitter power and source-rate allocation that preserves the existing TCP stack for rate-allocation. In Part II we focus on multicarrier systems. In particular, algorithms and protocols for dynamic spectrum management of digital subscriber line (DSL) access networks are developed. We derive a protocol called SCALE that jointly optimizes the power spectrum of all DSLs in a distributed manner having low complexity implementation. These developments give rise to new water-filling techniques devoid of bisection or channel sorting steps, that are particularly useful to applications of iterative water-filling. Lastly, novel algorithms for band preference design are developed to enhance the performance of current-day DSL networks.