`
`Book
`
`November 28, 2000
`
`13:19
`
`1
`Ad Hoc Networking
`An Introduction
`
`Charles E. Perkins
`Nokia Research Center
`
`In recent years, mobile computing has enjoyed a tremendous rise in popu-
`larity. The continued miniaturization of mobile computing devices and the
`extraordinary rise of processing power available in mobile laptop comput-
`ers combine to put more and better computer-based applications into the
`hands of a growing segment of the population. At the same time, the mar-
`kets for wireless telephones and communication devices are experiencing
`rapid growth. Projections have been made that, by the year 2002, there
`will be more than a billion wireless communication devices in use, and
`more than 200 million wireless telephone handsets will be purchased an-
`nually. The rise of wireless telephony will change what it means to be “in
`touch”; already many people use their office telephone for taking messages
`while they are away and rely on their mobile telephone for more impor-
`tant or timely messages. Indeed, mobile phones are used for tasks as simple
`and as convenient as finding one’s associates in a crowded shopping mall
`or at a conference. A similar transformation awaits mobile computer users,
`and we can expect new applications to be built for equally mundane but
`immediately convenient uses.
`Much of the context for the transformation has to do with keeping in
`touch with the Internet. We expect to have “the network” at our disposal
`for the innumerable little conveniences that we have begun to integrate into
`our professional lives. We might wish to download a roadmap on the spur
`of the moment so that we can see what is available in the local area. We
`might wish to have driving suggestions sent to us, based on information from
`the global positioning system (GPS) in our car, using the services offered
`by various web sites. The combination of sufficiently fast and inexpensive
`wireless communication links and cheap mobile computing devices makes
`this a reality for many people today. In the future, the average traveler is
`likely to take such services for granted.
`
`1
`
`Apple Inc. 1017
`U.S. Patent No. 9,445,251
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`2
`
`1 AD HOC NETWORKING
`
`Today we see a great expansion in the production of technology to
`support mobile computing. Not only are the computers themselves getting
`more and more capable, but many new applications are being developed
`and wireless data communications products are becoming available that
`are much improved over those available in the past. The bandwidth now
`available to laptop computers over radio and infrared links is easily 10 to
`100 times more than that available just ten years ago.
`Such rapid technological advance has spurred equally impressive growth
`in mobile connectivity to the Internet. In the wired Ethernet domain, we
`have plug-and-play hardware and software so that laptop computers can be
`reconnected with ease according to the form factors of the local network
`outlets. The Internet is available around the world to those willing to make
`a dial-up connection to a local phone number. People are getting used to the
`advantages of having frequent and convenient Internet access. As a result,
`more and more network functionality will be taken for granted by typical
`laptop users.
`As wireless network nodes proliferate and as applications using the
`Internet become familiar to a wider class of customers, those customers
`will expect to use networking applications even in situations where the
`Internet itself is not available. For instance, people using laptop computers
`at a conference in a hotel might wish to communicate in a variety of ways,
`without the mediation of routing across the global Internet. Yet today such
`obvious communications requirements cannot be easily met using Internet
`protocols. Providing solutions to meet such requirements is the subject of
`this book. The proposals to be described allow mobile computer users with
`(compatible) wireless communication devices to set up a possibly short-
`lived network just for the communication needs of the moment—in other
`words, an ad hoc network.
`At the same time, there is a huge potential market for embedded net-
`work devices in our vehicles, our mobile telephones, and perhaps even in
`our toys and personal appliances. Surely the day is not far off when a typi-
`cal child’s doll will have a microprocessor and a remote control device and
`will depend on network access to interact with the home’s television and
`computer games. Embedded networking could represent the “killer app” for
`wireless networks.
`Anyone reading this book will agree that the modern age of network-
`ing represents one of the great achievements of humanity. We already take
`many aspects of it for granted. In particular, we often take for granted the
`infrastructure currently needed to support our vast networking enterprise.
`The things we do with our networks do not inherently depend on the net-
`work infrastructure; rather, having the infrastructure extends the reach of
`network applications immeasurably.
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`Sec. 1.1. Model of Operation
`
`3
`
`Once we have grown accustomed to the power of network communi-
`cations and to accomplishing our daily tasks with the aid of applications
`that rely on networking, we will want the applications to be available at all
`times. In fact, many network researchers predict that some day in the not
`too distant future we will put our applications to use “anytime, anywhere,”
`perhaps by way of the rapidly expanding satellite communications systems
`now under construction. The communications satellites girding the earth
`will complement the cellular (wireless) telephone infrastructure, which is
`itself growing even more rapidly in most developed countries.
`Indeed, the authors of this book suggest that mobile computers and
`applications will become indispensable even at times when and at places
`where the necessary infrastructure is not available. Wireless computing de-
`vices should physically be able to communicate with each other, even when
`no routers or base stations or Internet service providers (ISPs) can be found.
`In the absence of infrastructure, what is needed is that the wireless devices
`themselves take on the missing functions.
`In this introductory chapter, we consider some general topics that pro-
`vide context for the rest of the chapters in this book. In the next sec-
`tion, we describe a general model of operation for ad hoc networks and
`some of the factors affecting the design decisions that various approaches
`have taken. In Section 1.2, we list a few of the commercial opportunities
`that may await vendors of wireless products when the necessary proto-
`cols are available. This will naturally include a look at some of the ap-
`plications enabled by ad hoc networking. Following that, Section 1.3 will
`discuss some of the technical drivers for the resurgence of interest in ad
`hoc networking. The needs of military communications have been very in-
`fluential in creating this renewed interest. Discussion of military ad hoc
`networking, however, is not included in that section because it is covered
`much more completely in Chapter 2. Because many of the approaches to
`ad hoc networks use variations on existing routing protocols, some very
`general comments about routing protocols are presented in Section 1.4.
`Finally, a capsule summary of each chapter in the book is presented in
`Section 1.5.
`
`1.1 MODEL OF OPERATION
`This book is concerned with ways (past and present) that wireless mobile
`computing devices can perform critical network topology functions that
`are normally the job of routers within the Internet infrastructure. Keeping
`track of the connections between computers is something so basic that a
`computer network, almost by definition, cannot exist without it.
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`4
`
`1 AD HOC NETWORKING
`
`There are many kinds of protocols available today that are supported
`by network infrastructure, either in a particular enterprise or in the Internet
`at large. These other protocols deserve consideration, but need adaptation
`before they can be useful within a network no longer connected to the In-
`ternet infrastructure. Some of them may not be appropriate for use when
`the infrastructure is unavailable; credit card validation and network man-
`agement protocols come to mind.
`As a matter of definition, an ad hoc network is one that comes together
`as needed, not necessarily with any assistance from the existing Internet
`infrastructure. For instance, one could turn on 15 laptop computers, each
`with the same kind of infrared data communications adapter, and hope that
`they could form a network among themselves. In fact, such a feature would
`be useful even if the laptops were stationary.
`There are a bewildering variety of dimensions to the design space of ad
`hoc networks. We take a particular slice of that design space that should
`serve a large number of user requirements and yet allow discussion of a num-
`ber of interesting and illuminating techniques. Besides ad hoc networking,
`similar techniques have been proposed under the names instant infrastruc-
`ture [Bagrodia+ 1996] and mobile-mesh networking [SDT 1995].
`Consider, for example, whether the range of wireless transmission should
`be large or small compared to the geographic distribution of the mobile
`wireless nodes. If all of the wireless nodes are within range of each other,
`no routing is needed, and the ad hoc network is, by definition, fully con-
`nected. While this might be a fortunate situation in practice, it is not a
`very interesting routing problem to solve. Plus, the power needed to obtain
`complete connectivity may be impractical, wasteful of battery power, too
`vulnerable to detection, or even illegal.
`Thus, we discuss only proposals that offer solutions to the case in which
`some of the wireless nodes are not within range of each other. Combined
`with the lack of infrastructure routers, the restricted range of wireless trans-
`mission indicates the need for multihop routing.
`As another example, we might suppose that wireless computer users
`could measure their relative positions and subsequently configure their lap-
`top computers using the measured distances, so that the appropriate link
`information could be available at each mobile node. This would work, but
`it would not be very convenient. Worse yet, the link information would
`be likely to change whenever the users moved relative to each other. We
`are not interested in simplifying the problem space at the expense of user
`convenience, however, so we restrict our attention only to those proposals
`that provide automatic topology establishment (eschewing user configura-
`tion steps) and dynamic topology maintenance (enabling user mobility). In
`fact, we make the slight additional restriction of considering only proposals
`that are self-starting, except possibly for an enabling or mode setting step
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`Sec. 1.1. Model of Operation
`
`5
`
`performed by the user, who should be able to exert necessary controls over
`the performance of the ad hoc networking operation.
`In this book, most of the discussion focuses on the interesting cases
`that have the following characteristics:
`• The nodes are using IP, the Internet Protocol [Postel 1981a], and they
`have IP addresses that are assigned by some usually unspecified means.
`• The nodes are far enough apart so that not all of them are within
`range of each other.
`• The nodes may be mobile so that two nodes within range at one point
`in time may be out of range moments later.
`• The nodes are able to assist each other in the process of delivering
`packets of data.
`
`The discussion in this book focuses on the protocol engineering that under-
`lies the establishment of the paths by which the ad hoc network nodes can
`communicate with each other. Thus, address autoconfiguration in particu-
`lar, a very interesting subject, is largely absent from this book, but is ripe
`for exploration very soon.
`As an example of a small ad hoc network, consider Figure 1.1 (taken
`from Chapter 3), illustrating a collection of eight nodes along with the
`links between them. The nodes are able to move relative to each other; as
`that happens, the links between them are broken and other links may be
`established. In the picture, MH 1 moves away from MH 2 and establishes new
`links with MH 7 and MH 8. Most algorithms also allow for the appearance
`of new mobile nodes and the disappearance of previously available nodes.
`
`MH3
`
`MH4
`
`MH5
`
`MH2
`
`MH6
`
`MH8
`
`MH1
`
`MH7
`
`Figure 1.1. An Ad Hoc Network of Mobile Nodes
`
`MH1
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`6
`
`1 AD HOC NETWORKING
`
`1.1.1 Symmetric Links
`Some of the models considered in this book depend on the existence of
`symmetric communication links between the nodes in the ad hoc work.
`Unfortunately, wireless links in the real world do not necessarily conform
`to this assumption. This assumption of symmetry is made because routing
`in networks with unidirectional links is known to be quite difficult. Recent
`analysis has yielded a mathematical result of interest that characterizes this
`difficulty [Prakash 1999]. As it turns out, there are such networks in which
`two system-wide (i.e., all-node) broadcasts are needed for a source to find
`a route to a particular destination. On the other hand, not all networks
`with unidirectional links exhibit this characteristic. If the network has a
`sufficiently high degree of connectivity and relatively few unidirectional
`links, alternative routes comprising only symmetric links can usually be
`found.
`There is another factor that mitigates the decision to ignore possible
`asymmetric routes. A unidirectional link is sometimes on the verge of fail-
`ure anyway. In such cases, extending the basic ad hoc network protocol to
`deal with unidirectional links may cause less robust routes to be discov-
`ered, leading to early failure and the subsequent need for a new (and more
`complicated) route discovery cycle.
`
`1.1.2 Layer-2 Ad Hoc Solutions
`The ad hoc networking protocols in this book are mainly targeted at layer-3
`operation. It is possible, in practically every case, to retool the protocol for
`use at layer 2. Doing so requires first that IP address fields be enlarged to
`contain 48 (or more) bits instead of 32, as needed for IP, because the IEEE
`MAC address is typically 48 or 64 bits long. This is not a problem, especially
`in view of the fact that such retooling will be needed anyway to enable the
`protocols to handle ad hoc networks of IPv6 addressable computers.
`However, given the universal deployment of IP applications for net-
`working today, every application eventually causes the communication sub-
`system to resolve an IP address into a neighboring layer-2 address, or else
`into the layer-2 address of a node in the neighborhood that can forward
`the application protocol data units (packets) toward the IP address of the
`desired application endpoint. When the route table at the application’s
`source node has IP addresses of desired endpoints, IP forwarding naturally
`takes care of framing the data with a layer-2 header containing the layer-2
`destination address of the next hop along a path toward the destination.
`This happy circumstance evaporates, however, when the routing is
`based on layer-2 addresses. In that case, as a matter of consistent terminol-
`ogy, the route table is presumably indexed by layer-2 destination addresses.
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`Sec. 1.1. Model of Operation
`
`7
`
`That implies that the IP address of the destination has to be resolved to
`the layer-2 address of the destination, even for destinations that are mul-
`tiple hops away. Then, unless the layer-2 route discovery is equipped with
`suitable layer-3 address information in appropriate extensions, additional
`broadcast discovery operations will be needed. If the layer-3 information is
`included for better performance, the entire operation can be viewed as a
`layer-3 route discovery anyway, albeit with an odd data structure for the
`route storage.
`
`1.1.3 Proactive versus Reactive Protocols
`One of the most interesting aspects of recent investigations concerns whether
`or not nodes in an ad hoc network should keep track of routes to all possible
`destinations, or instead keep track of only those destinations of immediate
`interest. A node in an ad hoc network does not need a route to a destina-
`tion until that destination is to be the recipient of packets sent by the node,
`either as the actual source of the packet or as an intermediate node along
`a path from the source to the destination.
`Protocols that keep track of routes for all destinations in the ad hoc
`network have the advantage that communications with arbitrary destina-
`tions experience minimal initial delay from the point of view of the ap-
`plication. When the application starts, a route can be immediately se-
`lected from the route table. Such protocols are called proactive because
`they store route information even before it is needed. They are also called
`table driven [Royer+ 1999] because we can imagine that routes are available
`as part of a well-maintained table.
`However, proactive protocols suffer the disadvantage of additional con-
`trol traffic that is needed to continually update stale route entries. The ad
`hoc network we are trying to support is presumed to contain numerous
`mobile nodes. Therefore, routes are likely to be broken frequently, as two
`mobile nodes that had established a link between them will no longer be
`able to support that link and thus no longer be able to support any routes
`that had depended on that link. If the broken route has to be repaired,
`even though no applications are using it, the repair effort can be consid-
`ered wasted. This wasted effort can cause scarce bandwidth resources to
`be wasted and can cause further congestion at intermediate network points
`as the control packets occupy valuable queue space. Since control packets
`are often put at the head of the queue, the likely result will be data loss
`at congested network points. Data loss often translates to retransmission,
`delays, and further congestion.
`As a result, on-demand, or reactive, protocols have been designed so
`that routing information is acquired only when it is actually needed. Re-
`active routing protocols may often use far less bandwidth for maintaining
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`8
`
`1 AD HOC NETWORKING
`
`the route tables at each node, but the latency for many applications will
`drastically increase. Most applications are likely to suffer a long delay when
`they start because a route to the destination will have to be acquired before
`the communications can begin.
`One reasonable middle point between proactive and reactive protocols
`might be to keep track of multiple routes between a source and a destina-
`tion node. This “multipath routing” might involve some way to purge stale
`routes even if they are not in active use. Otherwise, when a known broken
`route is discarded, one of the other members of the set of routes may be
`attempted. If the other routes are likely to be stale, the application may
`experience a long delay as each stale route is tried and discarded.
`
`1.1.4 Multicast
`Ad hoc networks are interesting to a large extent because of the challenge
`of maintaining a communication path between a source and a destination,
`even when some of the intermediate forwarding nodes are unable to continue
`participating in packet forwarding and must be replaced by nodes along
`another path. It turns out that maintaining paths between a single source
`and multiple destinations is somewhat more difficult, but not excessively
`so. Given the growing importance of multicast as a means to reduce the
`bandwidth utilization for mass distribution of data, and the pressing need to
`conserve scarce bandwidth over wireless media, it is natural that multicast
`routing should receive some attention for ad hoc networks.
`It is an open question whether or not the multicast routing algorithm
`should be integrated with the routing algorithm used to establish commu-
`nication paths between single endpoints. On the one hand, the problems
`may be sufficiently different so that trying to make a single routing algo-
`rithm serve for both may be unnaturally difficult. On the other hand, the
`problem of reestablishing paths caused by movement of intermediate points
`in a routing path or tree may dominate both situations. In this latter case,
`careful attention to path reestablishment for both contexts at once may be
`much easier than re-examining solutions in multiple disparate contexts, as
`would be necessary if multicast routing were unrelated to unicast routing.
`
`1.2 COMMERCIAL APPLICATIONS
`OF AD HOC NETWORKING
`In this section, we look at some of the potential applications for ad hoc
`networks that might provide the basis for commercially successful products.
`In fact, any commercially successful network application can be considered a
`candidate for useful deployment with nodes that can form ad hoc networks.
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`Sec. 1.2. Commercial Applications of Ad Hoc Networking
`
`9
`
`For example, users of nodes in an ad hoc network are likely to wish to
`transfer electronic mail. If some nodes in an ad hoc network offer web
`service, the other nodes that wish to make use of the service will still need
`to connect to the web server and support the usual HTTP traffic.
`
`1.2.1 Conferencing
`Perhaps the prototypical application requiring the establishment of an ad
`hoc network is mobile conferencing. When mobile computer users gather
`outside their normal office environment, the business network infrastructure
`is often missing. But the need for collaborative computing might be even
`more important here than in the everyday office environment. Indeed, the
`whole point of the meeting might be to make some further progress on a
`particular collaborative project. Given that today’s project environments
`are heavily computerized, for projects in a very broad range of industries
`the need for being able to create an ad hoc network seems clear.
`As it turns out, the establishment of an ad hoc network for collaborative
`mobile computer users is needed even when there may be Internet infra-
`structure support available. This results from the likely overhead required
`when utilizing infrastructure links, which might entail drastically subopti-
`mal routing back and forth between widely separated office environments.
`Current solutions for mobile networking (e.g., Mobile IP [Perkins 1996])
`are not well suited for efficiently supporting ad hoc networks, although the
`techniques are not wholly incompatible.
`This interplay between the immediacy of ad hoc networks and the pos-
`sible performance loss inherent in relying on Internet infrastructure routing
`is a recurring theme that should be considered in connection with each
`approach described in the chapters of this book.
`
`1.2.2 Home Networking
`As another example, consider the scenario that will likely result if wire-
`less computers become popular at home. These computers will probably be
`taken to and from the office work environment and on business trips. It is
`quite possible that such computers will not have topologically related IP
`addresses, especially if they are connected at the offices of each parent or at
`the children’s school. Keeping in mind the convenience that an unchanging
`IP address affords to the user, it would be nice to allow the various mobile
`computers to operate an ad hoc network in the home, even if the home main-
`tains its own subnet with more or less permanently situated network nodes.
`Add to this the fact that assigning multiple IP addresses to each wire-
`less node for identification purposes would add an administrative burden
`(a job that most people do not want), and the alternative of deploying an
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`10
`
`1 AD HOC NETWORKING
`
`ad hoc network (automatically created as needed) seems more attractive.
`Ad hoc networking offers the prospect of reachability to all the nodes at
`home regardless of their “normal” point of attachment, which would other-
`wise be indicated by the network prefix that is part of every IP address.
`Furthermore, by using protocols such as Mobile IP, the nodes in the home
`ad hoc network can operate as if they were still connected to their standard
`computing environment in addition to participating (with higher perfor-
`mance) in the residential ad hoc network.
`
`1.2.3 Emergency Services
`When created at home or away from home at a meeting, an ad hoc network
`makes up for the lack of an existing Internet infrastructure. But, what about
`cases in which the existing infrastructure is damaged or out of service for
`other reasons? We are all familiar with situations in which loss of local
`power causes loss of electricity, and each year natural disasters wreak havoc
`with people’s lives around the world. As the Internet grows in importance,
`the loss of network connectivity during such natural disasters will become
`an ever more noticeable consequence of the calamity. Furthermore, network
`applications will become increasingly important for emergency services, and
`thus it will be important to find ways to enable the operations of networks
`even when infrastructure elements have been disabled as part of the effects
`of a disaster.
`Ad hoc networks can help to overcome network impairment during dis-
`aster emergencies. Mobile units will probably carry networking equipment
`in support of routine operations for the times when the Internet is avail-
`able and the infrastructure has not been impaired. With the techniques
`and protocols in this book, emergency mobile units can greatly extend the
`usefulness of their networking equipment during times of lost infrastruc-
`ture support. For instance, police squad cars and firefighting equipment
`can remain in touch longer and provide information more quickly if they
`can cooperate to form an ad hoc network in places not otherwise offering
`connectivity to the global Internet.
`
`1.2.4 Personal Area Networks and Bluetooth
`The idea of a personal area network (PAN) is to create a very localized
`network populated by some network nodes that are closely associated with
`a single person. These nodes may be attached to the person’s belt or carried
`in a purse. More exotic visions of the future include virtual reality devices
`attached around the head and other devices more oriented toward the sense
`of touch. These devices may or may not need to have an attachment to the
`wide area Internet, but they will almost certainly need to communicate
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`Sec. 1.2. Commercial Applications of Ad Hoc Networking
`
`11
`
`with each other while they are associated with their users’ activities. In
`this scenario, mobility is not the overriding consideration.
`However, mobility suddenly becomes much more important when in-
`teractions between several PANs are needed. In other words, when people
`meet in real life, their PANs are likely to become aware of each other also.
`Since people usually do not stay in a fixed location with respect to each
`other for very long, the dynamic nature of this inter-PAN communication
`should be obvious. Methods for establishing communications between nodes
`on separate PANs could benefit from the technologies of ad hoc networks.
`No current discussion about PANs would be complete without at least
`some mention of Bluetooth [Haartsen 1998]. Bluetooth is an emerging short-
`range radio technology targeted at eliminating wires between personal dig-
`ital assistants (PDAs). If each PDA is equipped with a Bluetooth radio, it
`is possible for up to eight devices to organize themselves into what is called
`a piconet, with slotted communications controlled by a master. Bluetooth
`protocols at the physical and MAC layers are focused on saving battery
`power, as PDAs are much more useful if people do not have to be continu-
`ally changing dead batteries.
`When there are more than eight devices, Bluetooth requires that mul-
`tiple piconets be formed. These piconets can be connected together into a
`scatternet if one of the slaves agrees to relay data between two of the mas-
`ters. This suggests that the slave should have separate time slots in each
`piconet to reduce latencies for data transfers. It may even be necessary to
`form scatternets for the PAN that is associated with a single person. It is
`almost guaranteed that scatternets will be required for the interaction of
`multiple PANs.
`
`1.2.5 Embedded Computing Applications
`The world is full of machines that move, and future intelligent mobile ma-
`chines will be able to process a great deal more information about the
`environment in which they operate. The “environment” itself will increas-
`ingly be a virtual one created by fixed and mobile computers. Some re-
`searchers [Weiser 1993] predict a world of ubiquitous computing, in which
`computers will be all around us, constantly performing mundane tasks to
`make our lives a little easier. These ubiquitous computers will often react
`to the changing environment in which they are situated and will themselves
`cause changes to the environment in ways that are, we hope, predictable
`and planned.
`Many of these intelligent machines will be both mobile and connected
`by wireless data communications devices. The Bluetooth short-range radio
`device is expected to cost less than $5 within four years and to be incor-
`porated into millions of wireless communications devices. Already, many
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`12
`
`1 AD HOC NETWORKING
`
`computers and PDAs are equipped with inexpensive wireless ports. These
`are being used for synchronizing data between machines owned by the same
`person, for exchanging virtual business cards, for printing out small files on
`local printers, and so on.
`Ubiquitous intelligent internetworking devices that detect their en-
`vironment, interact with each other, and respond to changing environ-
`mental conditions will create a future that is as challenging to imagine
`as a science fiction scenario. Our world will change so much that it is
`hard to predict the kinds of applications that might predominate. Security
`considerations must be taken into account, of course, to prevent unwar-
`ranted intrusions into our privacy and to protect against the possibility of
`impersonation by other people. However, these security matters are well
`known in other contexts, and they are not the particular subject of this
`book.
`These capabilities can be provided with or without the use of ad hoc
`networks, but ad hoc networking is likely to be more flexible and conve-
`nient than, say, continual allocation and reallocation of endpoint IP ad-
`dresses whenever a new wireless communication link is established. Fur-
`thermore, once we become used to having these simple and easily imagined
`features, it seems a sure bet that new applications will be invented. In
`fact, we may become so enamored of ad hoc computing that we will begin
`to presume the presence of appropriate support environments. These envi-
`ronments would be made available to ad hoc mobile nodes and could be
`considered a new form of micro-infrastructure. We might normally expect
`our ad hoc computers to have access to local information about tempera-
`ture, light-switch controls, traffic information, or the way toward a water
`fountain [Hodes+ 1997]. Infrastructure elements that make their informa-
`tion available by way of standard TCP/IP client–server applications could
`operate in dual mode so that they participate in ad hoc networks as well,
`depending on the circumstances.
`
`1.2.6 Sensor Dust
`Recent attention has been focused on ideas involving the possibility of co-
`ordinating the activities and reports of a large collection of tiny sensor
`devices [Estrin+ 1999, Kahn+ 1999]. Such devices, cheap to manufacture
`and able to be strewn about in large numbers of identical units, could offer
`detailed information about terrain or environmental dangerous conditions.
`They might be equipped with positional indicators; alternatively, positional
`information could be inferred for network information such as the number
`of hops between various sensors and a well-known data collection node.
`Such sensor network nodes have two characteristics that will strongly
`influence the design of networks to facilitate data acquisition:
`
`
`
`Perkins-47190
`
`Book
`
`November 28, 2000
`
`13:19
`
`Sec. 1.2. Commercial Applications of Ad Hoc Networking
`
`13
`
`• Once situated, the sensors remain stationary.
`• Population may be largely homogeneous.
`• Power is likely to be a scarce resource, so sophisticated communica-
`tions scheduling will be important (see Section 1.3.2 and Chapter 10).
`• In fact, the lifetime of the battery may define