Enhanced multi-user DMT spectrum management using polynomial matrix decomposition techniques

Adebayo, Patrick 2019. Enhanced multi-user DMT spectrum management using polynomial matrix decomposition techniques. PhD Thesis, Cardiff University. Item availability restricted.

Preview	PDF - Accepted Post-Print Version Download (3MB) | Preview
	PDF - Supplemental Material Restricted to Repository staff only Download (138kB)

Abstract

This thesis researches the increasingly critical roles played by intelligent resource management and interference mitigation algorithms in present-day input multiple output (MIMO) communication systems. This thesis considers the application of polynomial matrix decomposition (PMD) algorithms, an emerging broadband factorisation technology for broadband MIMO access networks. Present DSL systems’ performance is constrained by the presence of interference (crosstalk) between multiple users sharing a common physical cable bundle. Compared to the traditional static spectrum management methods that define their survival to the worst-case scenarios, DSM methods provides some degree of flexibility to both direct channel and noise parameters to improve evolvability and robustness significantly. A novel crosstalk-aware DSM algorithm is proposed for the efficient management of multi-user DSL systems. Joint power allocation procedures are considered for the proposed single-channel equalisation method in DSL access networks. This thesis then shows that DSM can also benefit overdetermined precoding-equalisation systems, when the channel state information (CSI) parameters call for a specific decision feedback criterion to achieve a perfect reconstruction. A reasonable redundancy is introduced to reformulate the original multi-user MIMO problem into the simplest case of power management problem. DSM algorithms are primarily applied to solve the power allocation problem in DSM networks with the aim of maximising the system attribute rather than meeting specific requirements. Also, a powerful PMD algorithm known as sequential matrix diagonalisation (SMD) is used for analysing the eigenvalue decomposition problem by quantifying the available system resource including the effects of the crosstalk and its parameters. This analysis is carried out through joint precoding and equalisation structures. The thesis also investigates dynamic interference mitigation strategies for improving the performance of DSL networks. Two different mitigation strategies through a decision feedback equalisation (DFE) criterion are considered, including zero-forcing (ZF) and minimum mean square error (MMSE) equalisers. The difference between ZF and MMSE equalisations is analysed. Some experimental simulation results demonstrate the performance of both ZF and MMSE equalisation under the DFE equalisation constraint settings. Model reduction on the MMSE equalisation is thus applied to balance the crosstalk interference and enhance the data-rate throughput. Finally, the thesis studies a multi-user MIMO problem under the utility maximisation framework. Simulation results illustrate that the power allocation of multi-user DSL transmission can be jointly controlled and the interference can often be mitigated optimally on a single user basis. Driven by imperfect CSI information in current DSL networks, the research presents a novel DSM method that allows not only crosstalk mitigation, but also the exploitation of crosstalk environments through the fielding of versatile, flexible and evolvable systems. The proposed DSM tool is presented to achieve a robust mitigating system in any arbitrary overdetermined multi-user MIMO environment. Numerical optimisation results show that the mitigation of crosstalk impairment using the proposed DSM strategy. The design and implementation of the proposed DSM are carried out in the environment of MATLAB.

Item Type:	Thesis (PhD)
Date Type:	Submission
Status:	Unpublished
Schools:	Engineering
Uncontrolled Keywords:	Spectrum Management; Digital Subscriber Line (DSL); Discrete Multi-Tone (DMT); Precoding Feedback Equalisation; Decision Feedback Equalisation; Polynomial Matrix Decomposition.
Date of First Compliant Deposit:	3 May 2019
Last Modified:	05 Aug 2022 01:39
URI:	https://orca.cardiff.ac.uk/id/eprint/122133

Actions (repository staff only)

Edit Item