POWER LINE COMMUNICATION SYSTEMS

Abstract

The remarkably increasing demand for communication systems has recentlyforced the research community to consider power line (PL) networks for datatransmission, which is commonly referred to as power line communications(PLC). In particular, this technology becomes more attractive in harsh wirelessenvironments where radio spectrum is scarce or/and propagation lossis high such as in underground structures and buildings with metal walls.PLC can support many applications such as home-networking, internet andsmart grid. More specifically, PLC is considered the backbone of smart grids,not only because no extra wiring installation is required, but also becausePLC is a through-grid technology which could reduce the reliance of the utilitycompanies on third party connectivity and, consequently, overcome manysecurity and privacy issues. On the other hand, since PLs were not designedfor data transmission, communication signals over such channels can degradetremendously. Contrary to many other communication channels, noise overPLs cannot be described as additive white Gaussian noise. It is rather categorizedbroadly into impulsive noise and background noise with the former beingthe most crucial element influencing PLC systems. With this in mind, thisthesis will primarily focus on studying and developing advanced techniquesand algorithms to reduce the severity of impulsive noise over PL channels.The contributions of this thesis are described as follows. Initially, a thoroughreview is provided to introduce and compare the challenges facing PLC,PL channel and noise modelling schemes, as well as some noise mitigationtechniques. Next, novel approaches are proposed, with different degrees ofeffectiveness and complexity, to reduce the effect of impulsive noise in orthogonalfrequency-division multiplexing (OFDM)-based PLC systems. Firstly,an adaptive hybrid preprocessing system is introduced to improve the performanceof the conventional hybrid approach. In this respect, closed-form expressions for the output signal-to-noise ratio (SNR), probability of missed detection and probability of successful detection are derived. Secondly, andunlike existing works which are entirely based on countering impulsive noiseat the receiver side, we show that if the OFDM signal is preprocessed at thetransmitter side in such a way to minimize the signal peaks, the noise cancellationprocess can be made more efficient at the receiver. This is accomplishedby applying a peak-to-average power ratio reduction scheme such as selectivemapping. A closed-form expression for the probability of blanking error ofthis system is derived. Thirdly, to eliminate the estimation requirement problemof the short-term noise statistics associated with the aforementioned approaches,we propose the dynamic peak-based threshold estimation (DPTE)method. This method relies on utilizing the OFDM signal peak estimates withwhich optimal blanking is achieved irrespective of the noise parameters. Inaddition, to realize DPTE, a look-up table algorithm with uniform quantizationis exploited and investigated in various system configurations.Furthermore, this thesis explores the performance of multi-carrier codedivisionmultiple access (MC-CDMA) systems over the multipath PL channelcontaminated with Middleton class-A noise for various spreadingcodes, namely, Pseudonoise, Walsh-Hadamard and orthogonal poly-phase sequences.Different nonlinear preprocessors are implemented and the correspondingperformance is evaluated in terms of the output SNR and symbolerror rate.

Date of Award	1 Aug 2016
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Emad Alsusa (Supervisor)