Power Line Communications: Theory and Applications for Narrowband and Broadband Communications over Power Lines

Power Line Communications (PLC) is a promising emerging technology, which has attracted much attention due to the wide availability of power distribution lines. This book provides a thorough introduction to the use of power lines for communication purposes, ranging from channel characterization, communications on the physical layer and electromagnetic interference, through to protocols, networks, standards and up to systems and implementations. With contributions from many of the most prominent international PLC experts from academia and industry, Power Line Communications brings together a wealth of information on PLC specific topics that provide the reader with a broad coverage of the major developments within the field.

• Acts as a single source reference guide to PLC collating information that is widely dispersed in current literature, such as in research papers and standards.
• Covers both the state of the art, and ongoing research topics.
• Considers future developments and deployments of PLC

• Language: English
• Publisher: Wiley
• Release date: Jul 22, 2011
• ISBN: 9781119956280

Book preview

Contents

List of Contributors

Preface

List of Acronyms

1: Introduction

References

2: Channel Characterization

2.1 Introduction

2.2 Channel Modeling Fundamentals

2.3 Models for Outdoor Channels: LV Case

2.4 Models for Outdoor Channels: MV Case⁷

2.5 Models for Indoor Channels⁸

2.6 Noise and Disturbances

2.7 Measuring Techniques

2.8 PLC Channel Emulation Tools

2.9 Reference Channels for Access Domain

References

3: Electromagnetic Compatibility

3.1 Introduction

3.2 Parameters for EMC Considerations

3.3 Electromagnetic Emission

3.4 Electromagnetic Susceptibility

3.5 EMC Coordination

3.6 EMC Regulation in Europe

3.7 Final Remarks

References

4: Coupling

4.1 Introduction

4.2 Filtering Basics

4.3 Transformer-Capacitor Coupler Design

4.4 Impedance Adaptation Concepts

4.5 Experimental Verification

4.6 Further Possibilities

References

5: Digital Transmission Techniques

5.1 Introduction

5.2 Modulation and Coding for Narrowband PLC Systems

5.3 Modulation and Coding for Broadband PLC Systems

5.4 Conclusion

References

6: Protocols for PLC Systems

6.1 Introduction

6.2 Broadband PLC Media Access Control Layer

6.3 Protocols for PLC Supporting Energy Management Systems

6.4 Internet Protocol Television Over PLC⁷

References

7: Industrial and International Standards on PLC-based Networking Technologies

7.1 Introduction

7.2 PLC Standardization by Industrial Alliances

7.3 International Standards on PLC-Networking Technology

7.4 ETSI and CENELEC Standards

7.5 International EMC Product Standardization

References

8: Systems and Implementations

8.1 Introduction

8.2 PLC Smart Grid Systems

8.3 PLC Broadband Access Systems

8.4 Multimedia PLC Systems

8.5 DC-PLC Systems³

8.6 PLC in Emerging Countries

References

9: Conclusions

Index

titlepage

This edition first published 2010

© 2010 John Wiley & Sons Ltd

Registered office

John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, United Kingdom

For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com.

The right of the author to be identified as the author of this work has been asserted in accordance with the Copyright, Designs and Patents Act 1988.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher.

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books.

Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought.

Library of Congress Cataloging-in-Publication Data

Power line communications: theory and applications for narrowband and broadband communications over power lines/editors, H.C. Ferreira … [et al.]

p. cm.

Includes bibliographical references and index.

ISBN 978–0-470-74030-9 (cloth)

1. Electric lines-Carrier transmission. I. Ferreira, H. C. (Hendrik C.)

TK5103.15.P695 2010

621.382-dc22

2009053133

List of Contributors

Pouyan Amirshahi

Shared Spectrum Section 2.4

Masoud Ardakani

University of Alberta Section 5.3.4

Inigo Berganza Valmala

Iberdrola Sections 8.2 and 8.3

Gerd Bumiller

iAd GmbH Sections 6.3 and 8.2

Francisco J. Cañete

Universidad de Málaga Section 2.5

Giulio Colavolpe

University of Parma Section 5.3.3

Klaus Dostert

University of Karlsruhe Sections 2.2.5, 2.3, 2.7, 2.8, 2.9 and 5.3.1

Hendrik C. Ferreira

University of Johannesburg Editor, Section 5.2.5

Dario Fertonani

Scuola Superiore Sant’Anna Section 5.3.3

Stefano Galli

Panasonic Section 2.2 (except 2.2.5), Section 7.3

Holger Hirsch

University of Duisburg-Essen Chapter 3

Halid Hrasnica

Eurescom GmbH Sections 6.2 and 6.3

Piet A. Janse van Rensburg

Walter Sisulu University Chapter 4

Masaaki Katayama

Nagoya University Section 2.6

Mohsen Kavehrad

The Pennsylvania State University Section 2.4

Michael Koch

Devolo AG Chapter 3, Sections 7.4 and 7.5

Lutz Lampe

University of British Columbia Editor, Section 6.3

Haniph A. Latchman

University of Florida Sections 7.2 and 8.4

Sunguk Lee

University of Florida Section 7.2

Maxim Lobashov

Vienna University of Technology Section 6.3

John Newbury

The Open University Editor

Vladimir Oksman

Infineon Technologies Section 7.3

Moisés V. Ribeiro

Federal University of Juiz de Fora Section 8.6

Alberto Sendin Escalona

Iberdrola Sections 8.2 and 8.3

Thomas Stockhammer

Nomor Research GmbH Section 6.4

Theo G. Swart

University of Johannesburg Editor, Section 5.2.5

Andrea M. Tonello

University of Udine Section 5.3.2

Daisuke Umehara

Kyoto University Section 5.3.3

A. J. Han Vinck

University of Duisberg-Essen Section 5.2 (except 5.2.5)

Eric R. Wade

University of Southern California Section 8.5

Lawrence W. Yonge III

Intellon Corporation Section 8.4

Preface

With this book we took on the challenge to cover most of the technical field of Power Line Communications (PLC) with wide-ranging contributions on selected topics. The scope of this book is thus uniquely wide, not only for a book on PLC, but also to our knowledge for any book in the general field of Telecommunications. The inspiration for this wide coverage came from a survey of the many papers contributed to the International Symposium on Power Line Communications from 1997. The reader will thus find information widely dispersed in the literature, including research publications, standards documentation and even trade literature. We have attempted a coverage of both techniques and information on which there is currently consensus, as well as a limited selection of promising ones still under investigation.

The goal of this book is thus to inform newcomers to the exciting field of PLC, to inspire further research and perhaps to contribute to future consensus. This book may also pave the way for future books focusing more deeply on perhaps just one individual subfield of the various subfields covered here.

During this ambitious project, we involved 31 technical contributors from 27 institutions and 11 countries. Coordinationwas a huge task. The editors would like to express their sincere thanks to all our contributors.

As stated, this book was inspired by the International Symposium on Power Line Communications, which since 2006 has been an IEEE conference sponsored by the IEEE Communications Society. Much material included in our book evolved from the proceedings of this conference (refer to http://www.isplc.org/docsearch).

The editors would thus like to dedicate this book to Professor A. J. Han Vinck from the University of Duisburg-Essen, Germany, for his contributions to PLC. The organization of the first International Symposium on Power Line Communications in 1997 at the University of Essen was one of his many leadership initiatives during his career.

Han Vinck (right) receives the 2006 IEEE International Symposium on Power Line Communications Achievement Award. Lutz Lampe (left) presents the plaque at the 2006 IEEE International Symposium on Information Theory in Seattle, WA, USA.

images/fpref_image001.jpg

List of Acronyms

1

Introduction

The Editors

Power Line Communications (PLC) is currently an emerging technology, consequently it is attracting much attention. Research in PLC, which was formerly only low key, has intensified since the mid-1990s, attracting researchers and engineering practitioners from universities, industry and utility companies.

With this book we set out to create a wide-ranging introduction to PLC, ranging from channel characterization, communications on the physical layer and electromagnetic interference, through protocols, networks, standards and up to systems and implementations. We attempted to collate in one document information widely dispersed in the literature, often in hard-to-read research papers, and also in some cases obtainable only in standards which may still be in draft form. Selected topics from accepted practices and procedures, as well as from ongoing research, are covered. We sincerely hope that this book will also stimulate further research into the interesting but difficult topic of PLC.

Let us start by reviewing the historical development and some important issues concerning the general application of PLC, making ample use of the encyclopaedic overview article [1]. PLC has been studied for many years, although it has never been in the main stream of communications research activities. Early work in the twentieth century, with the goal of switching in substations, metering and basic load control can be attributed to Swiss engineers. During World War II, some radio amateurs experimented with PLC, when their activities on the Radio Frequency (RF) spectrum were restricted. As early as June 1954, the American Institute of Electrical Engineers (AIEE) published a report: ‘Guide to Application and Treatment of Channels for Power Line Carrier’. (For an updated version refer to reference [2].) Since then, much research has been done. Interest increased during the 1980s and intensified especially during the 1990s. A significant body of work has now been published in the research literature, as evidenced by the many references listed in this book. In fact, a number of subsystems and full systems have also been available from vendors for several years now.

Electrical power lines are usually classified into the high (>100 kV), medium (1-100 kV) and low (<1 kV) voltage networks, with respectively increasing communications difficulties. (Note that the above voltages represent rather loose bounds on the effective values, measured between phases in a three-phase network.) The main thrust of the research into PLC has been focused on low voltage (LV) electrical power distribution networks, which have also geographically the widest spread, and which usually have the most convenient access within various buildings and structures. It should, however, be noted that these LV electrical networks turn out to be rather hostile and unusual channels, due to the fact that their design never involved communications aspects. Most of the material in this book is thus focused on developing communication systems for the LV network.

Historically, the utility organizations have been an important driving force behind the development of PLC. A primary motivation has been to do load management. This is usually achieved by selectively switching off, at time of peak demand, such devices as water heaters, which consume much energy on the demand side. Some countries have employed a ripple control system for this purpose. A ripple control system is a unidirectional system with low data rates that typically operates in the frequency band below 3 kHz, and has the disadvantage that it may require several megawatts for information transmission. PLC is a much more sophisticated method, requiring much less power to achieve load management as well as several other functions.

A second important motivation in the development of LV PLC systems has been to facilitate meter reading from a distance. Potentially this may include not only electricity meters, but also water, gas and temperature meters. Developments in this direction were started in the USA, where meter reader salaries were relatively high and electricity companies have not been allowed to charge their customers fixed monthly amounts, as was sometimes the case in Europe [3]. It is interesting to note that an English study [4] showed that a human meter reader achieves an average information rate of only about 1 bit/s! This is indeed very low compared to what is possible with modern PLC systems. The metering information, apart from automatic billing, may be used for additional customer functions, such as warnings when no pay has been received or even disconnecting nonpayers. In recent years some developing countries have installed prepaid electricity facilities, without which it would have been economically impossible to provide electricity to their many low-income users. Within prepaid electricity facilities, PLC plays an indispensable role.

Electrical utility companies may, furthermore, use PLC to shut off parts of the network in the event of danger, to gather user statistics, to transmit information to selected users, or to broadcast simultaneously to all users. Due to concerns about the worldwide energy crisis, rising fuel costs and global warming, electrical power utility companies are currently placing much emphasis on the concept of the ‘Smart Grid’. Here, digital communications will be the backbone. Many utilities prefer to have full control over their own communications infrastructure, for economic and strategic reasons. It will be cost effective to use PLC at least for some of the communications links in a smart grid.

Disaster recovery, whether from natural or manmade disasters, is also currently a serious consideration. In the event of longer term power cuts to cellular and wireless communications equipment, using PLC may become the only option over some critical links.

Home automation and intelligent buildings represent another growing application area. On the physical communications level, PLC provide a natural communications link for various devices, such as the sensors of an alarm system. On a higher systems level, several research investigations and product developments have focused on employing the LV network as a local area network for conveniently connecting many different computers or consumer electronic systems in the same building.

The feasibility of several simultaneous frequency multiplexed analog voice channels has early in the history of PLC been demonstrated on the High Voltage (HV) network. The earliest systems were fixed carrier, analogue systems. Carrier frequencies up to 500 kHz were sometimes used. With power levels up to 80 W, ranges of several hundred kilometers could be achieved without repeaters (see, e.g., references [5,6]). Several systems providing voice communications over the HV network currently still work on these principles.

During the last decade, a number of pilot installations in different countries demonstrated the possibility of PLC to carry telephony (voice traffic), within a confined geographical area. In this way local access to the national telecommunications network may also become possible. This approach is especially attractive in developing countries. In these countries it is sometimes beneficial to employ the same costly cable infrastructure for dual purposes. Of special interest are the remote or geographically isolated villages where cellular telephony may not be economically feasible.

The scope for investigating future innovative applications is still wide open. For example, the transmission of slow scan television images over long distances can in many countries be used for security purposes and for the monitoring of distant installations. In developing countries, such a video communications facility may also be used, in conjunction with an audio channel, to realize broadcasts containing information or educational content to users, including remote and isolated communities.

PLC is usually considered as a retrofit facility, i.e. the electrical power reticulation network has already been installed and an advantage is thus that there are no additional costs pertaining to cables and related infrastructure. Also, the power network has the advantage of being an independent communications network. The HV and Medium Voltage (MV) networks cover long distances. During the last decades, many electricity companies have set up a fibre optic network in parallel to the HV network, mainly for signaling purposes. Only a fraction of the capacity of this fiber optic network is typically used, and it could therefore very well be used to form an extended telecommunications network, together with PLC systems operating on the MV and LV networks. The geographic coverage of the LV network is usually very wide where human habitation exists, and access to this network can be simple.

On the other hand, as stated before, power lines represent a particularly difficult communications environment. The noise on the PLC channel has very unusual characteristics and the noise levels may be excessive. The cable attenuation at frequencies of interest to communications is usually very large. Repeaters may thus be needed to compensate for cable losses, and to bridge over distribution transformers. Standing waves on long cables may lead to nulls in the frequency response. Care should be taken to circumvent the potential problem of large valued capacitors or inductors that may arise when designing for Low Frequency(LF) work. Furthermore, electromagnetic compatibility problems arise when interfacing electronic circuits with electrical power lines. To compound matters, important channel parameters such as impedance and attenuation, as well as the noise levels, fluctuate with time and load in a very unpredictable way.

Electromagnetic interference, possibly inflicted by PLC equipment to other users of the RF spectrum, is of much concern. Consequently, countering this interference is an important research area. An imperative focus of the standards for PLC developed duringrecent years has indeed been to minimize this interference. Note that we can distinguish between compliance standards and enforcement standards. The latter can be seen as mandatory regulations. It has thus become practice to constrain the transmission parameters of the PLC channel by international or national enforcements standards. Perhaps the most important signaling parameters which are specified by regulations and standards, are the maximum transmitted power and the allowable frequency bands. These parameters are restricted in order to prohibit or limit the interference with other telecommunications services, and to prevent further pollution of the electromagnetic spectrum.

It is interesting to note that one of the first directiveswas publishedin 1974by the Deutsche Bundespost: the ‘Technische Richtlinie für TF-Funkanlagen für industrielle und gewerbliche Zwecke’ [7]. Briefly, this early directive made provision for 5 mW of transmitted power within one channel covering 30–146 kHz. Later standards for narrowband PLC, such as the widely accepted CENELEC 50065–1standard followed and extended this early standard. This was followed by standardization activities from industry, standards bodies and professional societies.

Usually, narrowband PLC refer to low rate digital communications utilizing the frequency band below 150 kHz in Europe (following the CENELEC standard: 3–148.5 kHz), and below 450 kHz in the USA and some far eastern countries. To achieve higher data rate communications, wider bandwidths and hence higher upper frequencies became necessary, as dictated by the fundamental theorems from information theory and communications. Wideband PLC thus utilizes a much wider frequency band, with lowest frequency being typically 1 MHz, and with highest frequency typically up to 30 or 60 MHz. The use of higher upper frequencies, even extending into hundreds of MHz, has also been investigated and implemented - regulations vary between countries.

The most importantfactor delaying wider installation and use of PLC, especially wideband communications, has been the rather slow development of standards acceptable to all participants. Agreement on emission limits has been a key obstacle. The progress made in recent years on standards is also reported in this book. It should be noted that in general the category of ‘wire-line systems’ includes not only PLC systems, but also asynchronous digital subscriber loops and cable modem systems. Some standardization and regulation activities address this broader context of systems.

Through the years, different names and acronyms have been used for this interesting research and application area in communications. These terms have sometimes been confusing and even nebulous. Some alternative terms used through the years seem to have been PLT to denote Power Line Telecommunications or Power Line Transmission and DLC to denote Distribution Line Communications (i.e. communications for the LV network). PLC has once been used to denote Power Line Carrier (i.e. communications for the HV network). On the other hand, there seems to be wider agreement on using the acronym BPL to denote Broadband PLC, i.e. high data rate PLC requiring a wideband channel.

In this book we prefer the term ‘Power Line Communications’ and acronym PLC to encompass the whole field of communications over LV, MV and HV lines, whether narrowband or wideband. Here, we follow the trend set by the most influential conference in this field, namely the annual International Symposium on Power Line Communications or ISPLC, which was initiated at the University of Essen in 1997, and which became an official conference of the IEEE Communications Society in 2006. ISPLC has served as a very important forum for, and catalyst of research into PLC. A data base containing the full papers of the earlier ISPLC conference proceedings was initially established at the University of British Columbia, and is currently available under the auspices of the IEEE Communications Society at reference [8]. It now also provides links to other papers published in IEEE journals and conferences since 1986, and also to papers published in journals by Wiley, Elsevier and Hindawi, etc. The large body of ISPLC papers published up to now provided an important foundation of necessary material and also an inspiration for this book.

The rest of this book is constituted as follows.

The characterization of power lines and power distribution networks as a transmission medium for digital communications is one of the very first and fundamental steps towards successful design and implementation of PLC systems. Chapter 2 thus provides an overview of the state of the art in channel characterization for PLC.

The electromagnetic compatibility environment associated with any wire-line service is a cardinal factor when deploying it. Chapter 3 introduces the electromagnetic effects associated with wire-line systems, especially the specific issues of PLC and the potential to cause interference to established communication services.

Chapter 4 considers the coupling of the communications signal to the power line network. This is an underinvestigated topic with very scarce coverage in the literature, yet it can potentially yield a large gain in signal-to-noise ratios.

Digital transmission, the core of any communications system, is investigated for PLC systems in Chapter 5. We describe several contenders for modulation and coding, for both narrowband and wideband PLC systems. Widely accepted signaling techniques, as well as recent research results on promising new signaling techniques, are reported.

Networks and protocols reside on the top layer of PLC systems and may interface between the PLC system and an unusual mix of users. New protocols need to be developed because of the unusual interference on the physical channel, and the topology of LV distribution networks, which is different from other telecommunications networks. In Chapter 6 we consider a few different scenarios, representative of some practical or potential applications.

All wire-line and telecommunication services are expected to coexist with all other services. While communication standards are developed throughout the world, some standards are best established for local conditions in a particular country or continent. Chapter 7 provides an overview of standards and standardization activities, some of which may still be evolving.

In Chapter 8, a number of PLC systems and implementation scenarios are discussed to illustrate the potential and the ability of PLC technology for commercially viable communication solutions.

Finally, in the conclusions in Chapter 9 we consider some issues related to the future development and deployment of PLC.

References

[1] H. C. Ferreira, H. Grove, O. Hooijen and A. J. H. Vinck, Power line communication, in Encyclopedia of Electrical and Electronics Engineering (ed. J. Webster), Wiley, 1999, pp. 706–16.

[2] Power System Communications Committee, Summary of an IEEE guide for power-line carrier applications, IEEE Trans. on Power Apparatus and Systems, PAS-99(6), 2334–7, Nov./Dec. 1980.

[3] S. J. Holmes and D. Campbell, Communicating with domestic electricity meters. Proceedings of the International Conference on Metering Apparatus and Tariffs for Electricity Supply, Manchester, UK, Apr. 3–5, 1990, pp. 129–33.

[4] B. E. Eyre, Results of a comprehensive field trial of a United Kingdom customer telemetry system using mainsborne signaling. Proceedings of the International Conference on Metering Apparatus and Tariffs for Electricity Supply, London, UK, Apr. 13–16, 1987, pp. 252–6.

[5] Westinghouse power line carrier telephone system. Westinghouse Electric Corporation, Relay and Telecommunications Division, Coral Springs, FL 33065.

[6] ABB ETL power line carrier system - The best of a long line. ABB Netcom Ltd, Power System Communications, CH-5300 Turgi, Switzerland.

[7] Technische Richtlinie für TF-Funkanlagen für industrielle und gewerbliche Zwecke, Sep. 1974, Deutsche Bundespost/Fernmeldetechnisches Zentralamt, Referat S24, FTZ 17 TR2022.

[8] PLC DocSearch. Available: http://www.isplc.org/docsearch/ [6 March 2010].

2

Channel Characterization

P. Amirshahi, F. Cañete, K. Dostert, S. Galli, M. Katayama and M. Kavehrad

2.1 Introduction

John David Parsons writes in the preface to the first edition of Mobile Radio Propagation Channel [1]: ‘Of all the research activities that have taken place over the years, those involving characterization and modeling of the radio propagationchannel are among the most important and fundamental.’ While this view is perhaps not fully unbiased, we need to agree with Parsons that ‘the propagation channel is the principal contributor to many problems and limitations that beset [mobile radio systems]’ Power Line Communications (PLC) systems. Reference to mobile radio systems is quite fitting here, as, like the wireless channel, power lines have also not been ‘designed’ for carrying communication signals and, as we shall see in this chapter, wireless and PLC channels share a number of characteristics important for the design and performance of communication systems.

In the spirit of Parsons’ remarks, we start our journey on PLC with the characterization of power lines and power distribution networks as transmission media for data communication. More specifically, this chapter is intended to provide an overview of the state of the art in channel characterization for PLC. In doing so, we pay particular attention to specific models for channel transfer functions in different PLC environments and the characterization of disturbances experienced in PLC.

This treatment is organized in eight parts. Section 2.2 sets the stage with an overview of basic power line topologies and characteristics of power line channels. Note that nearly all channel models available today fall into three main categories, namely deterministic, empirical and hybrid models, and the advantages and disadvantages of these approaches are discussed. Sections 2.3 to 2.5 are dedicated to channel characterization for different PLC environments, with a focus on models for the channel transfer function. Measurements and mathematical models for the noise experienced in PLC systems are presented in section 2.6. These models are markedly different and put more structure on the noise than is seen in the conventional additive white Gaussian noise model, which is the default model used in the communications community. All channel modeling requires an empirical basis, whether as a starting point or for verification. Therefore, section 2.7 presents a short treatment of techniques for measuring the transfer function of power line channels. This is followed by an overview of hardware emulation tools for power line channels suitable for the analysis and design of PLC solutions in section 2.8. Finally, section 2.9 provides a list of nine reference channels for broadband access PLC systems, which were defined within the Open PLC European Research Alliance (OPERA) project to enable reproducible performance results and comparisons between different PLC solutions.

2.2 Channel Modeling Fundamentals

Among the main technical challenges in power line communications, the power line channel is a very harsh and noisy transmission medium that is very difficult to model [2-5]. The power line channel is frequency-selective, time-varying and is impaired by colored background noise and impulsive noise. Additionally, the structure of the grid differs from country to country and also within a country and the same applies for indoor wiring practices.

Due to the difficulty of modeling the power line transfer function, the first modeling attempts were mostly based on phenomenological considerations or statistical analysis derived from extensive measurement campaigns. More recently, papers attempting deterministic approaches have been appearing, thus indicating that a more basic understanding of the physical propagationof communicationssignals over power lines is now emerging. It is remarkable that the results of these recent deterministic approaches actually confirm the validity of some of the conjectures formulated during the time that analytical approaches were not deemed feasible, e.g. the multipath nature of signal propagation along power line cables.

Another important feature of the power line channel is its time-varying behavior. The channel transfer function of the power line channel may vary abruptly when the topology changes, i.e. when devices are plugged in or out, and switched on or off. However, the power line transfer function exhibits a time-varying behavior even if the topology of the network and the load (appliances) attached to it do not undergo abrupt changes. In particular, the power line channel exhibits a short-term variation because the High Frequency (HF) parameters of electrical devices depend on the instantaneous amplitude of the mains voltage. In addition, the noise injected into the channel by appliances is also dependenton the instantaneousamplitude of the mains voltage. Therefore, a cyclostationary behavior on the time selectivity of the channel as well as on the noise arises, and the period is typically half the mains period. An example of this behavior unique to the power line channel is shown in Figure 2.1, where the measured time variation of an indoor power line channel transfer function and the noise waveform generated by a halogen light with dimmer are shown. Despite the importance and uniqueness of this behavior, there are very few contributions that address this characteristic (see references [6-10] and a recent contribution that specifically addresses the issue of mapping directly in the discrete time domain the modulated input signal to the output when the time variability of the channel is taken into consideration [11]). In reference [7], the deterministic and random input-output relation for a Linear Time Varying (LTV) system is found for power line channels both in the time and frequency domains. In reference [11], the convolution operator is expressed in matrix form and both the Linear Time Invariant (LTI) and the LTV cases are considered; in the LTI case the channel is modeled by usual Toeplitz matrices, while in the LTV case the channel is modeled by special banded matrices. The important aspect to point out is that traditional channel models based either on multipath propagation or on Transmission Line (TL) theory are static and fail to capture time selectivity.

Figure 2.1 (a) Measured time variation of an indoor power line channel. Time selectivity occurs at a period rate, whereas frequency selectivity exhibits correlation at various time instants. (b) Noise waveform created by a halogen light with dimmer over one 60 Hz mains cycle.

images/c02_image001.jpg

Several groups are pursuing methods to deduce relevant statistical behavior from ensembles of physical models and measurements; e.g., references [12-15]. Other groups are instead following a deterministic approach based on precise channel models, e.g., references [16-24]. Statistical models do not require knowledge either of the link topology or of the cable models, but require an extensive measurement campaign. Deterministic models require detailed knowledge of the link topology and of the cable models, but do not require measurements.

Recent results seem to indicate that, if properly modeled, the power line channel transfer function exhibits more determinism than is commonly believed. This determinism could be exploited for robust modem design and system optimization. For example, the symmetry of the power line channel (mathematically proven and experimentally validated in references [25] and [23]) opens the door to information-theoretic considerations on optimal transmission when the channel is known at the transmitter. For example, treating known intersymbol interference as interference known at the transmitter allows us to use more effectively precoding schemes such as Tomlinson-Harashima or, more generally, dirty paper coding. Moreover, as reported in reference [23], it is possible to isolate reflections and resonant modes on the basis of specific features of the power line topology. This property of superposition of resonant modes allows us to assess more effectively the similarities (correlation) between the power line transfer functions pertaining to the same home, and this knowledge can be embedded into the adaptive equalizer of a power line modem. It is also important to point out that, with the exception of the hybrid approaches described in references [21], [24] and [9], no currently available model is capable of embedding this intrinsic correlation.

2.2.1 Brief Review of Indoor/Outdoor Topologies

¹

Indoor and outdoor power line topologies differ greatly from country to country, and also within each country. The major characteristics of these two environments will be reviewed here, whereas more details will be given in sections 2.3, 2.4 and 2.5.

2.2.1.1 Low, Medium and High Voltage Mains Topologies

High voltage (HV) lines bear voltages in the 110–380 kV range and span very large geographical distances. These lines have been used as a communications medium for voice since the 1920s [26] via single-sideband amplitude modulation (power carrier systems). Nowadays, PLC over HV lines comprises both analog systems (tele-protection) and digital systems (voice and data transmission).

Typically, Medium Voltage (MV) (10-33 kV) and Low Voltage (LV) (100-400 V) power distribution lines are used for high-speed PLC communication. MV power systems are typically deployed in a loop configuration, but sometimes they can be found deployed as open-loop systems and tree systems with radial arranged lines. Additionally, distribution lines consist of either underground or overhead cables.

In a single-phase configuration, a hot and a return (neutral) wire are fed to the premises’ main panel. Sometimes a separate ground (earth) wire is also added. This configuration is typical of small residential buildings. Generally, the power company distributes three phases, and only one of these is fed to a house, whereas a neighbormay be served off another phase. In the USA,² voltage ratings are 60 Hz 120 V, allowing a range of 114–126 V (ANSI C84.1). The new harmonized nominal voltage in Europe³ is 230 V (range, 207–253 V) 50 Hz (formerly, 240 V in the UK, 220 V in the rest of Europe).

The two-phase configuration is not common in Europe, but is typical in the USA in the split-phase configuration. In a typical home in the USA three cables come into the premises panel from the service. A center-tapped step-down transformer is located on the electrical line pole with the tap grounded and each socket connected across one side of the transformer. Larger devices (electric stoves, central air conditioning units, electric dryers, etc.) are wired across the entire transformer, receiving 240 V. In the USA, sometimes apartment complexes are even fed with a 120/208 V ‘wye’ configuration. The transformer is set up in a ‘y’ configuration with 208 V between any two secondaries, and 120 V between any one secondary and the center-tapped neutral. In some areas, two legs of a 120/208 ‘y’ are fed instead of the usual 120/240 split phase-service.

Three-phase (three hot wires plus one return) configurations are common in Europe, but not in the USA. The three-wire system that the user sees is typically derived from three-phase distribution, which uses a four-wire or five-wire system. In the five-wire system, there are three hot wires, one neutral wire and one grounding wire. The common three-wire receptacle uses only one of the three hot wires. In Europe most use 230/400 V, where the 230 V can be found between any of the three phases and neutral and the 400 V can be found between two of the three phases. The phase difference between phases is 120 degrees.

2.2.1.2 Residential and Business Indoor Wiring Topologies

Tree or star configurations are almost universally used. In Europe, both two-wire (ungrounded) and three-wire (grounded) outlets can be found. Interestingly, if two- or threephase supply is used, separate rooms in the same apartment may be on different phases. The UK has its exceptions and uses special ring configurations; a single cable runs all the way around part of a house interconnecting all of the wall outlets and a typical house will have three or four such rings. Moreover, in the UK the return cable will generally be grounded at the local substation, not locally in the house. There are also some cases, especially in old buildings, where only two wires run around the house (neutral and ground share one wire).

Power cables used for single-phase indoor wiring are comprised of three or four conductors in addition to the ubiquitous earth ground. These include ‘hot’ (black), ‘return’ (white), safety ground and ‘runner’ (red) wires, all confined by an outer jacket that maintains close conductor spacing.

While the ‘white’ return wires and safety grounds are isolated throughout all distal network branches, many national and international regulatory bodies today mandate that the return (white) and ground cables be connected together or ‘bonded’ via a low resistance shunt RSB at the service panel as shown in Figure 2.2. There is substantial mode coupling created by the electrical path through RSB, so that the effects that ground bonding have on the transfer function of the channel are rather substantial [23,24]. As described in reference [27], let us consider the simple topology shown in Figure 2.3(a): a two-conductor link, i.e. hot and return wires only, with the far outlet terminated on a matched resistor. This configuration should exhibit a simple attenuation profile which increases with frequency and has no notches. We also consider the case where this topology is composed of three cables (hot, return, and ground), and its return and ground cables are bonded (see dashed line) at the main panel as shown in Figure 2.3(b). The measured transmission responses without bonding (top trace) and with bonding (lower trace) are compared in Figure 2.3(c). The upper trace shows a benign 3 dB attenuation over 0–30 MHz in the absence of bonding. The addition of bonding creates significant resonant attenuation at 3.27 and 9.95 MHz, with less pronounced attenuation at 16.89 and 23.27 MHz. This measurement confirms that ground bonding creates significant effects on the channel transfer function. Nevertheless, the modeling of grounding practices has always been neglected and it has been taken into account for the first time only recently by Galli and Banwell [17,22,24].

Figure 2.2 Diagram of a typical service panel with one phase, four circuit breakers, and two branch cables shown along with two additional loads. The black (BLK) wires are each fed via separate circuit breakers while the white (WHT) wires connect to the mains transformer return (RTN) via a common terminal block. The safety grounds (GND) are connected to earth ground via a second terminal block. Element RSB represents a low shunt resistance between the ground and return paths, referred to as ‘bonding’. [22] © [2005] IEEE.

images/c02_image002.jpg

Figure 2.3 (a) Simple topology without ground bonding; (b) same topology as in (a) with ground bonding; (c) measurement of the effects of bonding on the transfer function from X to Y of the topologies in (a) and (b): without bonding (top trace) and with bonding (lower trace). [23] © [2005] IEEE.

images/c02_image003.jpg

As described earlier, wiring and grounding practices can be widely different and this makes modem design much more challenging. However, international harmonization has been underway for the past 20–30 years, and many regulatory bodies, such as the US National Electric Code (NEC), have revised and mandated a harmonized set of practices. In particular, the following practices are now mandatory in most parts of the world:

1. Typical outlets have three wires: hot, neutral and ground.

2. Classes of appliance (light, heavy duty appliances, outlets, etc.) must be fed by separate circuits.

3. Neutral and groundare separate wires within the home, exceptfor the main panelwhere they are bonded.

Although complex network topologies can indeed exist, the above-mentioned regulations greatly simplify the analysis of signal transmission over receptacle circuits.

2.2.2 Some Fundamental Definitions and Properties of Band-Limited Channels

Within the bandwidth of the channel W, let us define the channel frequency response H (f) as:

(2.1) images/c02_image113.jpg

where |H(f)| is the amplitude response characteristic and θ(f) is the phase response characteristic. The group delay of the channel (also called as the ‘envelope delay characteristic’) is:

images/c02_image171.jpg

A channel is said to be ‘ideal’ or distortionless if the following conditions are both satisfied:

(a) The channel amplitude response is constant: |H(f)| = H, ∀f ≤ W.

(b) The channel group delay is constant: τ(f) = τ, ∀f ≤ W. This is also equivalent to saying that the channel phase response θ (f) is a linear function of frequency.

If only condition (a) is satisfied, then the channel creates delay distortion. If only condition (b) is satisfied, then the channel creates amplitude distortion. If both conditions (a) and (b) are satisfied, then the constant value t of the group delay is also the propagation delay of the channel.

It is interesting to look at the behavior of the group delay near a channel notch. In this regard, it is useful to express the group delay as

(2.2)

images/c02_image114.jpg

where ℑ[z] represents the imaginary part of complex number z.

Let us now recall the following property of the continuous Fourier transform:

images/c02_image172.jpg

Using the above property, we can now write

images/c02_image173.jpg

where ℜ[z] denotes the real part of complex number z, and

images/c02_image174.jpg

where FT{·} denotes the continuous-time Fourier transform operator. From (2.2), it can be seen that the group delay of a channel can become very large at those frequencies where the channel amplitude response vanishes, i.e. in correspondence of channel nulls.

In the next subsections we will introduce some important channel metrics. In this regard, let us now define the following sequences:

Discrete time impulse response: {hi = h(t = iTS), i = 0, 1, 2,…, N − 1}, obtained sampling at rate FS = 1/TS the continuous time impulse response h(t). The channel memory is N − 1, and there are N nonzero taps.

Discrete frequency transfer function:

images/c02_image175.jpg

obtained as the N-point Discrete Fourier Transform (DFT) of the discrete impulse response hi.

2.2.2.1 Impulse Response Duration

Several definitions of this metric can be found in the literature, and often the channel impulse response duration is mistakenly labeled as ‘delay spread’ or ‘maximum delay spread’. It is generally accepted to define the impulse response duration as the time interval that contains a certain percentage of the total energy of impulse response; typical percentage values are 99%, 99.9% and 99.99%. The truncation of the impulse response is necessary in the cases of measurements as noise contamination is substantial in power lines, and some optimal methods are described in reference [28]. If the impulse response is generated according to a model, then truncation of the impulse response may be avoided.

2.2.2.2 Average Channel Gain

The power line channel is frequency selective, so average channel gain images/c02_image104.jpg is calculated by averaging over frequency:

images/c02_image176.jpg

where the right-hand side has been obtained using the Parseval theorem.

2.2.2.3 Root Mean Square Delay Spread (RMS-DS)

The RMS-DS is defined as the square root of the second central moment of the power (or energy) delay profile. If TS is the sampling time and N TS is the duration of the eventually truncated impulse response, the RMS-DS στ can be expressed as follows:

(2.3) images/c02_image115.jpg

where the following relationships hold:

(2.4) images/c02_image116.jpg

(2.5)

images/c02_image117.jpg

The RMS-DS is usually much smaller than the impulse response duration, sometimes as much as an order of magnitude, although this is seldom pointed out in the literature. Note that µ0 and σ0 are the average delay and the RMS-DS normalized to a unitary sampling time. Quantities µ0 and σ0 are sometimes used in the literature (see, for example, reference [28]), but it is important to remember that they have to be scaled appropriately by TS as in (2.4).

2.2.3 Characteristics of the Indoor Channel in the HF and VHF Bands

A typical indoor power line impulse response, frequency transfer function and group delay plot are plotted in Figure 2.4. It is interesting to confirm the theoretical considerations made in the previous sections. For example, it can be seen that: the average group delay is close to the propagation delay; the delay spread is much smaller than the impulse response duration; the group delay has peaks at those frequencies where the amplitude of the transfer function vanishes.

Figure 2.4 Typical measured indoor power line channel: (a) impulse response; (b) frequency transfer function; (c) group delay versus frequency (average group delay is plotted as a bold straight line).

images/c02_image004.jpg

Generally, indoor power line channels in the HF band (2-30 MHz) exhibit the behavior shown in Figure 2.4. The frequency transfer function exhibits HF selectivity. Moreover, group delay also shows high variability with multiple peaks. This means that the power line channel exhibits both amplitude and delay distortion. This can be explained taking into account that bridged taps - which together with ground bonding account for most multipath generation have lengths that are comparable with the wavelength λ of HF signals (10 m ≤ λ ≤ 150 m). A bridged tap with a length equal to a quarter of λ causes a pi-shifted reflection to add coherently with the main signal, thus producing a notch at the frequency corresponding to wavelength λ. If the bridged tap length is a small multiple of the quarter wavelength, then the pi-shifted reflection will be only slightly attenuated with respect to the main signal and this will cause a dip in the frequency transfer function. Since bridged taps with lengths equal to a small multiple of the quarter wavelength of the frequencies in the HF band are very common for indoor topologies, the indoor power line channel in the HF region is characterized by many frequency notches and dips and, thus, by many group delay peaks as predicted by (2.2).

Several published papers report that typical delay spreads in homes are in the order of a few microseconds. For example, papers [29] and [30] report that the measured indoor delay spreads are in the order of 2–3 µs, with some exceptional cases of delay 5 µs. However, it is not really clear how the delay spread has been calculated and if the delay spread reported is actually the RMS-DS defined in (2.3) since the RMS-DS is often confused with the impulse response duration. This confusion on what metric is actually being reported as delay spread is common to most papers on power line channel characteristics.

The measurements reported here pertain to a set of channels measured by the HomePlug Powerline Alliance that have been recently made available online [31]. The set contains 120 power line channel impulse responses (both in forward and reverse directions), representing data from six different homes of varying sizes and age in North America. Additional details can be found in references [30] and [32].

The impulse response duration on these measured channels was calculated to include up to 99.9% of the impulse response energy. Based on 99.9% energy, RMS-DSs of the channels were calculated and they were found to vary between 0.1 µs and 1.7 µs. The plot of the RMSDS Cumulative Distribution Function (CDF) is plotted in Figure 2.5 (only the 60 channels in forward direction have been considered). Among the 120 measured responses, median value was found to be around 0.5 µs and 118 channels exhibited an RMS-DS below 1.31 µs. Two responses only exhibited a much higher RMS-DS: 1.73 µs and 1.81 µs, respectively. This confirms that the RMS-DS of the indoor power line channels is actually much smaller than usually believed. This has an impact on the choice of the parameters of a multicarrier system designed to operate over the power line channel.

Figure 2.5 Empirical CDF of the RMS-DS of the measured indoor power line channels. For more details, see reference [33] © [2009] IEEE.

images/c02_image005.jpg

The empirical CDFs of the channel gain images/c02_image253.jpg can be easily calculated on the basis of the measured data. The channel gain CDF can also be used to calculate the CDF of the link signal-to-noise ratios (SNRs) available at the receiver. The SNR available at the receiver is a random variable that can be expressed as a scaling of images/c02_image104.jpg :

images/c02_image177.jpg

where PTX and PN are the transmit and noise density powers, respectively. Galli reported in reference [33] for the first time that average and individual channel gains of measured indoor power line links are lognormally distributed. The probability that a link experiences an SNR larger than a certain amount γ can be expressed using the Complementary CDF (C-CDF) of the random variable SNR:

images/c02_image178.jpg

The resulting empirical C-CDF (probability that a link has an SNR higher than γ) is plotted in Figure 2.6 as a dashed curve, assuming a transmit power of PTX = −55 dBm/Hz and a noise floor of PN = −120 dBm/Hz. If the gain images/c02_image253.jpg is fitted to a normal distribution, then the SNR is also normal and its C-CDF is the one shown in Figure 2.6 as a solid curve. The C-CDF plotted in Figure 2.6 confirms that SNR in power lines is generally low. For example, the median SNR is equal to 15 dB and the probability that the link SNR is above 30 dB is only 8%. Similarly to channel gains, Galli reported for the first time that also the RMS-DSs of measured indoor power line links are lognormally distributed [33].

Figure 2.6 C-CDF of channel SNRs: empirical (dashed) and simulated (solid) when images/c02_image104.jpg is fitted to a lognormal distribution. For more details, see reference [33]. © [2009] IEEE.

images/c02_image006.jpg

Although there is today a growing interest in utilizing the VHF part of the spectrum for power line communications, very little data has been published on this matter. The characteristic of the power line channel in the VHF band (30-300 MHz) is very different from the one described above for the HF region, but has not been studied as in depth as for the HF case and very few experimental results have been reported in the literature. In particular, amplitude and delay distortion diminish considerably so that the channel RMS-DS is much smaller in the VHF band compared to the HF band. This can be explained taking into account that indoor bridged taps usually have lengths that are much longer than the wavelength of the VHF frequencies (1 m ≤ λ ≤ 10 m). Therefore, bridged tap lengths are typically a large multiple of the quarter wavelength so that the pi-shifted reflection will be heavily attenuated with respect to the main signal and this will cause just a small dip in the frequency transfer function. Therefore, the indoor power line channel in the VHF region is characterized by much fewer and less pronounced dips than in the HF case. As a consequence, delay distortion will also be small as fewer and less pronounced group delay peaks would be present. On the other hand, attenuation also increases but not as much as one would imagine. The only paper to report a quantitative analysis of channel characteristics between 30 MHz and 100 MHz was published in 2006 by Schwager et al. [34]. A very interesting result reported there was that the median channel attenuation in the 30–100 MHz range was only 4 dB higher than in the HF band (see Table I of