Continuing in the spirit developed in the previous chapter, rather than looking at meshes by pursuing a linear layer-by-layer exposition of the protocol stack as in Figure 3.1, we will continue to take a more pragmatically integrated view. This chapter and the next chapter therefore look at two key aspects of mesh systems, or indeed of any communications system; these are susceptibility to interference and quality of service. PHY, MAC, routing, transport and application behaviours along with their interactions are all relevant, although this chapter on susceptibility is more related to the lower layers and Chapter 6 on the quality of service is more related to the higher layers.

We begin by looking at interference and how the mesh may react to it. We do this by firstly classifying all the various forms which interference may take.

At the physical layer the effect of interference depends on the modulation and coding in use within the mesh. Of course this is true of any communications system, but we find an important distinction is that a mesh precludes the easy use of some common modulation approaches. The reason for this is the typical lack of any centralised control within a mesh, which precludes approaches demanding synchronisation of modulation across nodes. Examples include many versions of frequency hopping.

At the MAC, the effect of interference depends on the MAC scheme in use. Once again this is true for any communications system, yet again we find an important distinction is that a mesh precludes the easy use of many common MAC approaches. This includes the common slotted schemes of FDMA, TDMA and CDMA.

As we have seen from the previous chapters, there are numerous key considerations to bear in mind when planning to implement a mesh. Some of these key considerations, if not properly addressed, constitute potential pitfalls for the mesh system designer. The aim of this short chapter is to bring all such considerations together for easy reference, so the pitfalls may be avoided. This is particularly appropriate as not all pitfalls have familiar equivalents outside the world of mesh networking.

In summary, potential pitfalls already covered in the body of this book centred around

1. capacity,

2. infrastructure,

3. efficiency,

4. relay exhaustion,

5. initial roll-out,

6. upgradeability,

7. reliance of the system on user behaviour, and

8. ad hoc versus quality of service.

There are also two areas which we have covered implicitly, but now wish to highlight explicitly here:

1. 9. security and trust, and

2. 10. system economics.

Let us deal with these areas in turn.

Capacity

In Chapter 4 we noted that it was often rumoured that meshes self-generate capacity, as if this were a truism. The reasoning behind such a claim was usually along the lines of ‘each new user brings additional capacity to the mesh’, or ‘each new user effectively becomes a base station’. This book critically examined such statements and separated the reality from a something-for-nothing type of mythology. We outlined the difference between network capacity and the user throughput which is actually available, concluding that user throughput cannot grow as fast as the mesh grows. The simple reason is the relay requirement imposed on each node, due to the traffic of other nodes.

Introduction

We are by now well acquainted with the ‘observe, decide, act’ cycle. This chapter is the first of three core chapters that relates to the cycle and focuses on the act part. ‘Taking action’ was introduced in the last chapter as ‘the setting of the various knobs on the radio’. So we have already been introduced to the idea that there are a large number of possible knobs such as frequency, bandwidth, signal duration, modulation technique, power, etc. that can be set, but we have not looked at any details. In this chapter we look at the details and more explicitly at what actions are needed for the kinds of applications described in the opening chapter of this book.

To do this we need first of all to further build our knowledge about what actions are possible. A second important point of this chapter is to develop an understanding of the consequences of the actions taken. While we have stressed in the previous chapter that ‘taking action’ is not just about the physicality of the transmitted signal and can pertain to other aspects of the communication process such as higher-layer performance issues, management of battery lifetime of the node or the processing resources of the node, it is the physical interaction of the transmitted signal with other entities around it that is core to understanding the consequences of the actions that are taken.

Introduction

This book has considered the essential issues associated with cognitive radio. Cognitive radio is a truly interdisciplinary topic. It crosses the fields of information theory, propagation studies, RF design, telecommunications, wireless networking, signal processing, artificial intelligence, cognitive science, software engineering, regulatory policies, security, application design, plus many many more. On the one hand the need for such a breadth of knowledge seems daunting and on the other hand it seems very exciting and opens up the way for new possibilities. In this last chapter an attempt is made to summarise the main points of the book.

A brief summing up

The first main point of the book is that a cognitive radio is not just a radio for dynamic spectrum access. The second is that dynamic spectrum access is a great idea and in looking at dynamic spectrum access many new paradigms for dynamic behaviour of radios come to light. The third point is that there are many potential applications on the horizon for cognitive radio as is hopefully clear from the simple mindmap in Figure 9.1.

Throughout the book the observe, decide and act cycle was placed at the core of the cognitive radio (Figure 9.2 recaptures the cycle). And secure, build and regulate were seen as the key additional activities of focus. Each of these topics warrants a few words.

The Internet is now firmly part of our everyday life. We perform many common tasks on-line, such as banking, grocery and gift shopping and the purchasing of travel or cinema tickets. Plus we get a growing portion of our entertainment from on-line sources: entertainment and social networking are two of the largest growth areas. We have seen the beginning of basic quality video from, for example, You Tube and the development of social networking sites such as My Space and FaceBook, which have been enormously popular, especially amongst younger generations of consumers. If we are to continue in this trend of doing more on-line, our need for bandwidth will increase. And in future we might expect to generate appreciable content ourselves, for upload onto the Internet, as well as to continue to download content. But that is not all; our need for Internet availability and quality will also increase.

It would be very convenient if such future Internet access were also wireless, with the near ubiquitous service we are used to from cellular phones. However, building a new network to achieve this, or upgrading an existing network to support this, would mean installing or changing a great deal of infrastructure. What then if a method existed which promised improved Internet access with fewer requirements for new infrastructure? This is widely advertised as the domain of the mesh network.

This chapter begins with a top-level introduction to mesh networking, then looks at how meshes may fit into the larger telecommunications infrastructure, before moving on to classify and explain the basic properties of a mesh.

In this chapter we will show that a startling effect in meshes is that quality of service (QoS) is not under the operator's control but depends on mesh node behaviour. In a mobile mesh, this means that your QoS depends on your neighbours' behaviours at any point in time, potentially spanning a range all the way from having no discernable effect up to a complete loss of your service. There is nothing quite like this problem in the networks we commonly use today.

But we begin this chapter by looking at how QoS is defined and what QoS levels are required for the applications of today and into the future. Following this we look at whether there are any useful services which truly only a mesh could support. After considering node mobility and showing how node to node relative speed is the key parameter, we look at an example of how a mesh may break into disconnected pieces. This can occur before the full mesh capacity is approached. Finally we show that mesh quality of service is not entirely within the control of the network operator, but rather depends on user mobility and traffic, before showing how adding infrastructure can help improve the quality of service position. Mitigation techniques for QoS issues induced by normal user activity include the provision of extra network-owned nodes in order to regain some control, but this comes at a cost for the operator.

This book has so far focused on meshes for telecommunications, however another use of multi-hop networking is the wireless sensor network. In fact it is potentially beginning to look like WSNs might outstrip telecommunications as a use for multi-hop and mesh technology. We have kept this chapter separate as WSNs have some unique properties, but we find that many aspects of mesh discussed earlier in the book apply to WSNs in much the same ways. In terms of applications, at the time of writing, smart buildings (advanced control of lighting and HVAC, heating ventilation and air-conditioning) and logistics look like the top two likely WSN applications, in terms of earliest uptake.

Let us begin with an introduction to wireless sensor networks. We take quite a broad overview before concentrating on the networking aspects of WSNs.

The role of a wireless sensor network is essentially that of a monitor. Broadly speaking, what is being monitored can usually be placed within one of three groups:

• area monitoring – i.e. monitoring somewhere; examples include the environment or area alarms (intrusion etc.);

• entity monitoring – i.e. monitoring something; examples include a civil structure (bridge, building etc.) or a human body;

• area–entity interaction monitoring – i.e. monitoring something, somewhere, in context; examples include vehicles on the road, asset tracking or the flow of a manufacturing process.

As to why a sensor network is important, it is most simply understood by realising that, often, individual sensors themselves are limited in their ability to monitor a given situation.

Overview

We listed the characteristics of an ad hoc mesh in Chapter 3, Table 1.1, and we built upon this to create the access mesh concept. But we have not so far attempted to offer any detailed explanation of the key mesh characteristics. The function of this chapter is to examine these fundamentals as a final foundation before Chapter 4, where we begin the detailed testing of the four key hypotheses of mesh performance which we introduced at the end of Chapter 1.

A logical way to address the fundamentals is to consider, in turn, each layer of a generic communications protocol stack as shown in Figure 3.1.

At the bottom of the stack is the physical layer, or PHY. This consists of the parts which directly concern the air interface, for example the antennas and transceiver electronics. By implication this also includes detail design elements, such as the choice of modulation scheme and transmit power.

But it does not include the method by which access to the air interface is determined – this is the job of the medium access control layer, or simply MAC. This, for example, will include schemes to allow multiple users to share the medium in some more or less fair fashion, such as the random collision avoidance approaches used in 802.11 or the structured time and frequency division multiplexing as used in GSM.

To enable nodes to find and communicate with each other, some sort of addressing scheme is required; this is contained in the routing layer. An example is the increasingly ubiquitous Internet protocol.

Introduction

The chapters so far have looked at the main functionality of a cognitive radio through an exploration of the ‘observe, decide, act’ cycle. We now step back from this and look at security issues specifically related to cognitive radio. This chapter is the shortest chapter of the book. This is not indicative of the level of importance of the topic of security and cognitive radio but of the fact that cognitive radio security has to date received much less attention than other topics.

All communication systems need to be made secure to operate. Typically any users of a system have to authenticate themselves on the network. Authentication is the process of determining whether someone or something is, in fact, who or what it is declared to be. Some authentication processes may involve the simple use of a password but others are more complex. The use of digital certificates, issued and verified by what is known as a Certificate Authority (CA) as part of a public key infrastructure, is an example of a more stringent process. Following authentication, authorisation processes, to ensure data and services are accessible only to those who have the correct entitlements, are needed. During all communication privacy may need to be guaranteed. Encryption is typically used to achieve this either using a public key infrastructure, a symmetric cryptographic approach or hash algorithms. Eavesdropping by man-in-the-middle attacks must be avoided.

Wireless mesh networking is a hot and growing topic, still in its infancy in some ways, whilst already shown to be capable in others. From a military beginning, mesh networks moved to civilian use and are now being deployed worldwide as both local area networks (LANs) and metropolitan area networks (MANs). However, these deployments are still ‘leading edge’ and it is not yet clear what the most enduring applications of mesh will be – particularly as the market moves from early adopters towards widespread take up.

Some of the claims for what a mesh network may deliver have been very ambitious to say the least. In this book we investigate such claims versus the real qualities of mesh networks and identify the key time scales and drivers for the challenges involved with making meshes. Throughout the book we attempt to keep mathematics to a minimum. Where an equation is shown, it remains practical to follow the flow of the book without needing to understand the maths fully.

The book takes a very pragmatic but balanced approach to the issues. We are particularly interested in meshes with an external access capability, for example to the Internet. We supply a technical assessment of mesh and multi-hop networking, highlight the attractions, identify the pitfalls, provide clear and concise hints and tips for success – summarised inside the back cover – and finally evaluate some real-world examples of good mesh applications. These include wireless cities, community networking and vehicular ad hoc networks (VANETs). Wireless sensor networks (WSNs) are another important application of mesh techniques with their own unique challenges, and these receive their own chapter.

Here we examine two of the most common real-world mesh deployments: firstly wireless cities and secondly community Internet. We show how their reasons for success align with the content presented in earlier chapters in this book. Interestingly, wireless city deployments are targeted at urban areas which already have wired Internet connectivity but where the addition of mobility is valued, whilst in contrast community Internet is targeted at those places where the wired Internet is sparse and connectivity can be added most easily by using wireless to serve fixed locations.

Thirdly, we also show a rising application of mesh networking – vehicular ad hoc networks (VANETs). These systems are targeted at improving road safety and have had spectrum allocated in many countries, and enjoyed success in industrial trials. We expect VANETs to experience particularly strong future growth.

Wireless cities

Several wireless cities are now up and running which provide easy Internet access on the move. In the UK, London and Bristol were early examples, whilst in the USA there is New York, Portland, OR and a rapidly growing number of others. The aim in each case is to enable easy mobile connection to the Internet. This can serve the general public, business users and the city authorities, who may use it for operational purposes, including for public services such as law enforcement.

The wireless nodes are deployed at street level and each includes a normal WiFi access point, so that users may connect with their existing WiFi enabled devices, such as laptops and a growing number of converged cellular-WiFi mobile handsets and PDAs.

To summarise once more, at this point in the book it has been shown that practical mobile meshes are not chosen primarily for spectral efficiency nor for any notion of self-generation of capacity. Meshes should be chosen because they have other benefits. Section 2.2 provided an introduction to how meshes offer coverage benefits, which is possibly their major attribute. In this chapter we revisit our six most likely applications which we have been considering throughout the book. These are

• cellular multi-hopping or WiFi hotspot extension,

• community networking,

• home and office indoor networking,

• micro base station backhaul,

• vehicle ad hoc networks (VANETs), and

• wireless sensor networks (WSNs).

The first five applications are considered in detail in this chapter, whilst wireless sensor networks receive their own treatment in Chapter 10, since they have some unique features. In this chapter, we also look at the barriers to mesh adoption and the time scales likely for them to be overcome.

For the following discussion we find it useful to group the applications into those which form a mesh on the user side and those which form a mesh on the network side, in other words those where the users' nodes themselves mesh together, versus those where only the backhaul forms a mesh. There is one case where the mesh can be for both users and network backhaul; this occurs in VANETs.

Introduction

Cognitive radio is a topic of great interest and holds much promise as a technology that will play a strong role in communication systems of the future. This book focuses on the essential elements of cognitive radio technology and regulation. This is a challenging task in that cognitive radio is still very much an emerging technology. There is much debate over its exact definition, its potential role in communication systems, whether cognitive radios should in fact be permitted in the first place and if yes, what the regulatory policies should be. However, while acknowledging the flux in this field, the book aims to identify the core concepts that will remain central to the field irrespective of how precisely it develops. The aim of this first chapter is to briefly define cognitive radio and to then focus on the all important question of why cognitive radios are needed. This chapter therefore motivates all that is to come in the book.

Brief history and definition

The term cognitive radio was coined by Mitola in an article he wrote with Maguire in 1999 [1]. In that article, Mitola and Maguire describe a cognitive radio as a radio that understands the context in which it finds itself and as a result can tailor the communication process in line with that understanding.