1468
`
`PROCEEDINGS OF THE IEEE, VOL. 66, NO. 11, NOVEMBER 1978
`
`Advances in Packet Radio Technology
`
`Invited Paper
`
`computer network resource sharing. It is well known that the
`computer traffic generated by a given user is characterized by
`a very low duty cycle in which a short burst of data is sent or
`received followed by a longer quiescent interval after which
`additional traffic will again be present. The use of dedicated
`circuits for this traffic would normally result in very ineffi-
`cient usage of the communication channel. A packet of some
`appropriate size is also a natural unit of communication for
`computers. Processors store, manipulate, and transfer data in
`fiiite length segments, as opposed to indefinite length streams.
`It is therefore natural that these internal segments correspond
`to the computer generated packets, although a segment could
`be sent as a sequence of one or more packets. Computer re-
`source sharing techniques which exploit the capabilities inher-
`ent in packet communications are still primarily in the research
`stage, but significant progress has already been achieved in this
`area [6].
`individual per-
`switching concept incorporates
`The packet
`packet processing by each switch or node in the network such
`that incremental network capacity can be dynamically allocated
`by a node immediately after it receives a packet. Each packet
`wends its way from node to node through the network until
`eventually it arrives at the final destination and is delivered.
`The transit time through the network is typically a fraction of
`a second. Due to the low duty cycle of individual user traffic,
`allocating a portion of the network capacity for that, user in
`advance (eg., to provide a dedicated circuit) could aIso lead to
`very inefficient use of the network resources even if the alloca-
`tion is valid over a very short time.
`An essential attribute of any network is its ability to provide
`full connectivity among all network participants. Full con-
`nectivity implies that any set of computers can communicate
`subject only to the overall network performance and adminis-.
`trative limitations. Specific performance parameters or con-
`straints such as throughput, delay, cost and reliability are usu-
`ally quite important
`to the critical mass of
`interconnected
`subscribers. Within a given network environment, delay,
`throughput, and cost are intricately related because lower de-
`lay usually means higher data rates which, in turn, implies
`higher throughput and greater cost. Of course, the delay can
`never be reduced below the speed of light propagation delay.
`Packet radio [7] is a technology that extends the
`original
`packet switching concepts which evolved
`for networks
`of
`point-to-point communication lines to the domain of broad-
`rapid development in this area has
`cast radio networks. The
`been greatly stimulated by the need to provide computer net-
`work access to mobile terminals and computer communica-
`tions in the mobile environment. Packet radio offers a highly
`efficient way of using a multiple access channel, particularly
`with mobile subscribers and
`large numbers of
`users with
`
`A I. INTRODUCTION
`
`N EXCITING set of developments has taken place during
`the last few years in the field of digital radio networks.
`The advantages of multiple access and broadcast radio
`channels for information distribution and computer communi-
`cations have been established and several experimental digital
`radio networks are now in operation. Packet radio is a perfect
`example of the rapid technological progress which has been
`achieved. It utilizes packet-switched communications and is
`in the
`particularly important for computer communications
`In this paper, we pro-
`ground mobile network environment.
`vide an overview of the basic concepts of packet radio and dis-
`cuss the recent technology and system advances in this field.
`We also address the closely related subjects of signaling in the
`ground mobile radio environment and the advantages of spread
`spectrum technology in the multiple access network environ-
`ment. This paper is intended to be tutorial with emphasis on
`the current state-of-the-art in packet radio.
`The rapid growth in packet communications which has taken
`place during the last decade following the successful develop-
`ment of the ARPANET [ 1 ], [ 21 is directly related to the in-
`creasing demand for effective telecommunication
`service to
`handle computer communications [ 31, [4]. Only in this time
`frame did it become cost effective to utilize minicomputers,
`and later microprocessors as packet switches in a large scale
`network [ 51 . In a packet-switched network, the unit of trans-
`mission is called a packet. It contains a number of data bits,
`and is usually of variable length up to a maximum of a few
`A packet includes all the addressing and
`thousand bits.
`control information
`necessary to correctly route it
`to its
`destination.
`Packet switching was originally designed to provide efficient
`network communications for “bursty” traffic and to facilitate
`
`Manuscript received July 1,1978;rcvised August 8,1978.
`R. E. Kahn is with Advanced Rasearch Projects Agency. U.S. Depart-
`ment of Defense, hlington. VA 22209.
`S. A. Gronemeyer is with Rockwell International, Dallas, TX.
`J. Burchfiel is with Bolt Beranek and Newman, Inc., Cambridge, MA
`02138.
`R. C. Kunzelmm is with SRI International, Menlo Park, CA 94025.
`
`U.S. Government work not protected by U.S. copyright
`
`Petitioner Emerson's Exhibit 1002
`Page 1 of 31
`
`

`
`KAHN et al.: ADVANCES IN PACKET RADIO TECHNOLOGY
`
`1469
`
`bursty traffic. During the early l 970's, the ALOHA project
`
`
`
`the transmissions are very short. The use of computer control
`
`
`
`
`
`
`at the University of Hawaii demonstrated the feasibility of
`
`
`
`
`for channel access can lead to very efficient system operation
`using packet
`
`
`
`broadcasting in a single-hop system (see reference
`
`
`
`relative to other more conventional manual methods of access
`
`
`[ 8], [ 9 J ). The Hawaii work led to the development of a mul­
`
`control [ 11 ] .
`
`
`
`tihop multiple access packet radio network (PRNET) under
`In recent years, the subject of efficient spectrum utilization
`
`
`
`
`
`
`
`
`
`the sponsorship of the Advanced Research Projects Agency
`
`
`
`
`has received increasing attention. A special issue of IEEE
`
`
`(ARP A). The PRNET is a fundamental network extension
`on spectrum
`
`
`Transactions on Electromagnetic Compatibility
`
`
`of the basic ALOHA system and broadens the realm of packet
`
`
`
`management [ 12] addresses this topic in considerable detail.
`
`
`communications to permit mobile applications over a wide
`
`
`This subject is not addressed further in this paper, other than
`
`
`geographic area. The use of broadcast radio technology
`
`
`
`to note that because of its capability for dynamic allocation of
`
`
`
`for local distribution of information can also provide a
`
`
`
`the spectrum, packet radio is a particularly good choice to
`
`
`degree of flexibility in rapid deployment and reconfigura­
`
`
`
`
`obtain efficient utilization for bursty traffic.
`
`The ability to
`
`
`
`achieve effective usage of the spectrum will be a central factor
`
`
`tion not currently possible with most fixed plant installations.
`
`
`along with cost in determining the ultimate viability of radio
`
`
`
`
`Although the original impetus for packet radio development was
`
`
`based networks for local distribution of information.
`
`
`
`
`
`and still is largely based on tactical military computer com·
`
`
`
`In this paper, we discuss the basic concepts of packet radio
`
`
`
`munication requirements [IO] , the basic concept is applicable
`
`
`
`
`and present the recent technology and system advances. In
`
`
`to an extremely wide range of new and innovative computer
`
`
`
`Section II, we indicate various capabilities and services which
`
`
`
`communication applications never before possible in any prac­
`
`
`
`a packet radio network might provide. In Section III, we
`
`tical way.
`
`
`
`
`consider the problem of signaling over a ground radio channel
`
`
`In addition to the strong ARP ANET and ALOHA system
`
`
`
`
`with all its attendant environmental factors. In Section IV,
`
`
`
`ip.fluences, three technical developments in the early l 970's
`
`
`
`
`we discuss the basic operation of a packet radio network with
`
`
`
`
`were largely responsible for the evolution of packet switching to
`
`
`
`underlying emphasis on the network elements and system
`
`
`the radio environment. The first was the microprocessor and as­
`
`
`In Section V, we discuss several advanced system
`protocols.
`
`
`sociated memory technology which made it possible to incorpo­
`
`
`
`capabilities for operation and control of the packet radio
`
`
`
`
`rate computer processing at each packet radio network node in
`
`
`
`system. In Section VI, we separate out a subject of particular
`
`a form that was compatible with mobile usage and portable
`
`
`
`interest, namely spread spectrum transmission in the network
`
`
`
`operation. The second was the reduction to practice of sur­
`
`
`environment. Although a packet radio system need not
`
`
`face acoustic wave (SAW) technology which can perform
`
`
`employ spread spectrum, there are several noteworthy
`
`
`
`
`matched filtering (to receive wide-band radio signals) on a very
`
`
`
`
`attributes arising from its use. The experimental packet radio
`
`
`small substrate of quartz or similar piezo-electric material.
`
`
`
`network (PRNET) under development by the ARP A is dis­
`
`
`
`
`The third development was purely conceptual and involved an
`
`
`
`
`cussed in Section VII. The final section contains conclusions.
`
`
`
`awareness within the computer and communications commu­
`
`
`
`
`nities of the importance of "protocols" in the development of
`
`
`network management strategies. It was upon those three pil­
`
`II. CAPABILITIES AND SERVICES OF A PR NE1WORK
`
`
`lars that the technical approach to the PRNET was founded.
`
`
`
`
`A primary objective of a packet radio network is to support
`
`
`Packet radio network technology will be essential for
`
`
`
`real-time interactive communications between computer
`
`
`military and other governmental needs as terminals and com·
`
`
`
`
`resources (hosts) connected to the network and user terminals
`
`
`
`
`puter systems become pervasive throughout essentially all
`
`
`
`
`(e.g., terminal-host, host-host, and terminal-terminal interac­
`
`
`aspects of their operations. Initially, the needs for radio-based
`
`
`tions). In order to satisfy this objective, the network should
`
`
`
`
`computer communications are expected to be prevalent in
`
`
`
`provide certain basic capabilities and services which can be
`
`
`training, on or near the battlefield, and for crisis situations. The
`
`
`grouped roughly into two categories: those which are always
`
`
`
`first operational systems, even if of limited availability, are
`
`
`
`or automatically provided by the network and those which
`
`
`
`most likely to be deployed for use in one of these areas where
`
`
`a user may select based on his application. The former cate·
`
`
`
`
`a higher relative cost of providing the advanced capability can
`
`
`
`gory includes such capabilities as network transparency, area
`
`
`
`. be tolerated. Within the civilian sector, there is also a strong
`
`
`
`coverage/connectivity, mobile operation, intemetting, coex­
`
`
`
`need for terminal access to information in the mobile environ·
`
`
`
`istence, throughput with low delay, and rapid deployment.
`
`
`ment, but the cost of service to the user (e.g., personal ter·
`
`
`
`
`The last category includes error control options, routing
`
`
`
`minal, tariffs) will dictate when such capabilities should be
`
`
`
`
`
`options, addressing options, and services for various tactical
`
`
`
`
`
`publicly provided. We expect to see a considerable increase in
`applications.
`
`
`the usage of civilian terminals and microcomputers "on the
`We identify here a few of these basic packet radio network
`
`
`
`
`move" during the early 1980's but, in contrast to the military
`
`
`
`environment, these applications are expected to involve
`
`
`
`services and capabilities. While the list is not intended to be
`
`
`
`
`relatively simple equipment, reduced capabilities and lower
`
`
`exhaustive, those items on it are all major factors of interest.
`costs.
`
`
`
`We assume that computer resources (hosts) need to be
`
`
`
`All users in a packet radio network are assumed to share a
`
`
`connected with each other and with individual users who
`
`
`
`common radio channel, access to which is controlled by micro·
`
`
`might access data bases, manipulate files, run programs or
`
`
`processors in the packet radios. In contrast to a CB radio
`
`write and execute programs to run on remote hosts. The
`
`
`
`channel in which contention for the channel is directly con·
`
`
`
`packet radio network merely provides a high throughput, low
`
`trolled by the users (who at best can do a poor job of schedul·
`
`
`delay means of interconnection for the (potentially mobile)
`
`
`ing the channel), the packet radio system decouples direct
`
`
`community of users. Many of these operations will be inter·
`
`
`
`access to the channel from user requests for channel access.
`
`
`
`active, with a computer response to a remote user entry being
`
`
`
`desired in real-time. Although the primary objective of the
`
`Within a fraction of a second, the microprocessors can dy­
`
`
`
`net is to provide service to computer communication traffic,
`
`
`
`namically schedule and control the channel to minimize or
`
`
`other types of service, such as might be required for real-time
`
`
`
`
`avoid conflicts (overlapping transmissions) particularly when
`
`Petitioner Emerson's Exhibit 1002
`Page 2 of 31
`
`

`
`1470
`
`
`
`PROCEEDINGS OF THE IEEE, VOL. 66, NO. 11, NOVEMBER 1978
`
`speech, can be accommodated along with the capability for
`
`
`
`of 0.1 s in nets of 100 mi area coverage size. Parameters on
`
`
`
`end-to-end security
`
`
`based on packet encryption techniques.
`
`
`
`this order are required to provide the real-time interactive
`
`
`
`services, and to accommodate efficient data transfers. With
`
`
`
`
`100 kbits/s signaling rates, a maximum packet size might be a
`A.Transparency
`
`few thousand bits.
`The basic internal operation of the network should be
`
`
`
`
`transparent to the user. We use this term to mean that all
`
`
`user data presented to the net should be delivered to the
`G.Rapid and Convenient Deployment
`
`
`
`
`destination without modification of the information content
`
`Deployment of the packet radio net should be rapid and
`
`in any way. Only the data to be delivered, and the necessary
`
`
`
`
`convenient, requiring little more than mounting the equipment
`
`control and addressing
`
`
`
`information should be required of the
`
`
`at the desired location. No alignment procedure should be
`
`
`
`delivery, user as input. All other aspects of routing, reliable
`
`
`
`required, and in most applications omni-directional antennas
`
`
`
`
`protocols and network operation should be handled by the
`
`
`would be used, thus eliminating the need for antenna align­
`
`ment. Once installed, the system should be self-initializing
`
`
`network itself. Only in the case of communication difficulties
`
`
`
`
`should the user or user process be advised of internal network
`
`
`
`and self-organizing. That is, the network should discover the
`
`
`
`radio connectivity between nodes and organize routing strate­
`
`
`
`status. Transparency is desirable in order to allow the net­
`
`gies on the basis of this connectivity and on the source/
`
`
`
`
`work to regulate and optimize its internal flow of traffic in a
`
`
`
`
`destination data of traffic presented to the net. Packet radios
`
`
`
`
`
`global manner without unnecessary constraints applied by
`
`
`
`should be capable of unattended operation.
`
`
`users. The users, in turn, need not be concerned with the
`
`
`
`activities taking place in the net or their effect on network
`
`
`operations, but need only specify the services desired.
`H.E"or Control
`Data integrity is crucial for most computer applications.
`
`
`
`
`
`
`Error control should be provided by the network, so that
`B.Area Coverage and Connectivity
`
`
`
`Area coverage with full connectivity should be provided.
`
`
`
`
`packets delivered to a user with undetected errors occur less
`
`
`
`For ground mobile radio, network diameters on the order of
`
`
`frequently than about one in 1010 packets. This is a critical
`
`
`100 miles are appropriate, but the system architecture should
`
`
`
`requirement for computer communications, since even one
`
`
`
`allow the geographic area of coverage to be expanded at the
`
`
`
`undetected error in a large file may render it useless or cause
`
`
`
`expense of increased end-to-end delay across the network.
`
`
`troublesome and unpredictable problems during subsequent
`
`
`
`All valid traffic originators within the net must be provided
`
`use of such a file. While detection of errors is essential,
`
`
`
`connectivity with all other valid receivers subject only to the
`
`
`choices exist in dealing with the detected errors. In some
`
`
`
`overall reliability and performance of the system. The net­
`
`
`
`cases, error detection and retransmission may be used, while
`
`work need have no prior knowledge of which users may wish
`
`
`
`in other environments, more sophisticated forward error cor­
`
`
`to connect to which other users or resources in the net. This
`
`
`
`rection technology must be used in order to maintain satis­
`
`
`
`is particularly important (and necessary) for the mobile
`
`factory throughput and delay when communicating through
`subscribers.
`land mobile radio channels.
`
`C.Mobility
`I.Routing Options
`The system should support mobile terminals and computers
`
`
`The network should support efficient communication
`
`
`
`
`
`at normal vehicular ground speeds within the area of coverage.
`
`
`between any pair of users and the capability for users to
`
`
`
`Packet radios in mobile applications must satisfy reasonable
`
`
`
`broadcast a packet to a subset or to all users on the net. Land
`
`
`size, weight, and power consumption constraints.
`
`
`mobile radio traditionally has been used for point-to-point
`
`
`voice communications. Dispatching services, walkie talkies,
`
`
`and, recently, CB radio all have supported a broadcast mode
`D.lntemetting
`
`
`
`of operation as well. These capabilities could be requested by
`The packet radio network structure should be capable of
`
`
`
`
`a "type of service" field provided by the user in the packet
`
`
`internetting in such a way that a user providing a packet
`header.
`
`
`address in another net can expect his network to route the
`
`
`
`associated packet to a point of connection with the other net
`
`
`
`or to an intermediate (transit) net for forwarding. Similarly,
`J.Addressing Options
`
`
`arriving internet packets should also be routed to the local
`The network should provide a capability for addressing a
`
`
`
`
`user.
`
`
`
`
`subset of the network participants and for efficiently establish­
`
`ing communication among them. This might be used for
`
`
`
`
`real-time conferencing or to support message delivery to a
`E.Coexistence
`
`
`list of addressees with the minimum ·number of network
`Radio frequency characteristics of the packet radio system
`
`
`
`
`
`
`transmissions. Logical network connectivity is a necessary
`
`
`
`should allow coexistence with existing users of a chosen
`
`
`
`foundation upon which protocols for these services can be
`
`
`frequency band. This could provide a greater degree of spec­
`built.
`
`
`
`trum sharing, particularly among similar systems, and may
`
`
`
`facilitate the introduction of the technology in new geographic
`locations.
`K.Tactical Applications
`
`In tactical military applications, the RF waveform used by
`
`
`
`
`
`
`
`
`packet radios should provide resistance to jamming, spoofing,
`F.Throughput and Low Delay
`
`
`
`
`
`detection, and direction finding. In many cases, waveforms
`
`The capacity of the packet radio system should allow for
`
`
`
`with these capabilities lead natural)y to the capability for
`
`
`
`
`variable length packet sizes up to a few thousand information
`
`
`
`
`
`position location and relative navigation. With the addition of
`
`
`
`
`bits, and provide delivery of packets with delays on the order
`
`Petitioner Emerson's Exhibit 1002
`Page 3 of 31
`
`

`
`KAHN e t al.: ADVANCES IN PACKET RADIO TECHNOLOGY
`
`1471
`
`a communication security function, the packet radio net
`would then provide an integrated communication, navigation,
`and identification system for secure tactical use.
`
`111. SIGNALING IN THE GROUND RADIO
`ENVIRONMENT
`is applicable to ground-based,
`Packet radio technology
`airborne, seaborne, and space environments. In this paper, we
`focus on ground-based networks which encounter perhaps the
`most difficult environment in terms
`of propagation and RF
`links, particularly when mobile
`connectivity. Ground radio
`terminals are involved, are subject to severe variations in
`strength due to local variations in terrain,
`received signal
`In addition, reflections give
`man-made structures and foliage.
`rise to multiple signal paths leading to distortion and fading
`as the differently delayed signals interfere at a receiver [ 131.
`As a result of these phenomena, RF connectivity is
`difficult
`to predict and may abruptly change in unexpected ways as
`mobile terminals move about. An important attribute of a
`packet radio system is its self-organizing, automated network
`management capability which dynamically discovers RF con-
`nectivity as a function of time for use in packet routing. The
`multipath phenomena also provide a strong motivating factor
`for the use of
`spread spectrum waveforms in packet radio
`In the paragraphs which follow, we first bound
`systems.
`(roughly) the radio frequency choices which are most appro-
`ground-based radio networks. We then discuss
`priate for
`the characteristics of propagation path loss, multipath effects,
`and man-made noise at these frequencies, and conclude with a
`discussion of spread spectrum signaling [ 141 and its applica-
`bility to ground-based packet radio systems.
`
`A . Frequency Band
`The operational characteristics of the radio frequency band
`have a major impact on the packet radio
`design. The lowest
`and highest frequencies which can be used for a packet radio
`system are determined primarily by considerations of band-
`width and propagation path loss (and the associated RF power
`generation requirement) respectively.
`the
`The required systems bandwidth effectively determines
`lowest desirable radio frequency in two
`ways. Practical,
`is difficult to achieve if the
`cost-effective radio equipment
`ratio of RF bandwidth to RF center frequency is much larger
`than about 0.3.
`This lower bounds the range of acceptable
`RF center frequencies. In practice, a center frequency
`well
`in excess of this lower bound is also desirable if the received
`signals would
`otherwise have too wide a multipath
`spread
`(e.g., due to sky wave phenomena at HF). For a packet radio
`system to deliver 2000 bit packets through a network with
`delays on the order of a tenth of a second, the data rate of
`the system must be in the range of a few hundred kilobits
`per second, which implies RF bandwidths of a few hundred
`kilohertz. From
`an implementation point of view, then, the
`be at least a few megahertz, or
`RF center frequency should
`in the lower high-frequency (HF) band extending from 3 MHz to
`30 MHz. Propagation in the HF band can provide long distance
`communication due to sky wave reflections from the earth’s
`ionosphere, but the propagation suffers from noticeable multi-
`path spreading of the signal which, as will be described later in
`the section, limits the data-rate of signals which can be used.
`Multipath spreading in the very*-frequency
`(VHF) band
`from 30 MHz to 300 MHz, where line-of-sight propagation
`
`to a few microseconds as
`dominates, is typically reduced
`compared to the millisecond spreads encountered at HF, and
`data rates on the order of a hundred kilobits can be supported.
`Multipath fading and distortion are still a problem at
`VHF,
`particularly for terminals which are mobile or do not operate
`with radio line-of-sight. However, diversity techniques or the
`spread spectrum signaling techniques discussed later can over-
`come these difficulties.
`The upper limits on usable radio frequencies for packet
`radio are primarily established by propagation path loss. As
`the operating frequency
`rises to about 10 GHz, absorptive
`losses due to the atmosphere and rain rapidly increase, and the
`resulting radio range is reduced accordingly. In general, packet
`radio systems must use closely spaced relays in order to pro-
`vide adequate area coverage at these frequencies. The cost of
`providing a dense relay population may be acceptable if the
`distribution of users is also dense and if packet radios co-
`users can provide the relay function. For
`located with the
`most applications, however, 10 GHz is a practical upper limit
`for a useful radio frequency in a ground-based packet radio
`system. We conclude, then,
`that practical packet radio
`sys-
`tems should use radio frequencies in the upper VHF band, in
`the ultra-high-frequency (UHF) band from 300 MHz to 3 GHz,
`and in the lower portion of the super high frequency (SHF)
`band from 3 GHz to 30 GHz.
`An additional factor which must be considered for opera-
`tion4 systems is the authorization to radiate packet radio
`transmissions. The VHF and UHF bands are already heavily
`spread spectrum signals potentially
`allocated. The
`use of
`could allow coexistence of a packet radio system with exist-
`ing users of some frequency band. However, this is a rela-
`tively new
`concept from the regulatory point
`of view, and
`significant technical issues would have
`to be resolved to
`establish the feasibility of coexistence.
`The discussion of propagation, multipath, and background
`noise which
`follows focuses on the UHF band, although,
`qualitatively, the phenomena
`discussed apply to the VHF
`and SHF bands as well. Later sections of the paper describe
`an experimental packet radio which operates at 1710-1850
`MHz in the upper UHF band.
`
`B. Propagation Characteristics
`Packet radio network operations would be greatly simplified
`radio line-of-sight path
`if all radios were sited such that a
`existed to nearby radios. Link design
`procedures for such
`paths are well understood, and RF connectivity within the
`network would then be fixed and reliable. Such stringent
`restrictions on siting are not reasonable from the user point
`of view, however. Many users of a packet radio network will
`have to operate from facilities previously established without
`consideration of radio propagation. Use of packet radio in a
`mobile environment would be almost useless if siting were
`required for reliable operation.
`loss is achieved on a radio
`The minimum theoretical path
`link in
`free space (i.e.,
`a vacuum), where received signal
`strength decreases as the inverse square of link range. For a
`the path loss of free space may be ap-
`ground radio link,
`proached on a link having a radio line-of-sight path, although
`even under this desirable condition diffraction and multipath
`phenomena can greatly reduce received signal strength.
`given
`When a radio
`line-of-sight path does not exist on a
`link, one can still speak of sited or non-sited terminals, due to
`
`Petitioner Emerson's Exhibit 1002
`Page 4 of 31
`
`

`
`1472
`
`the strong influence of shadowing by local terrain and objects
`and of the elevation of the antenna above the ground. A sited
`terminal is one which has been located to avoid surrounding
`obstacles and whose antenna has been elevated to the maxi-
`mum extent possible, while a terminal operating from a mov-
`ing vehicle would generally be nonsited.
`Average path attenuation exceeds that of a free space radio
`link by a significant amount in the ground radio environment,
`depending on the type of terrain and the elevation of the radio
`antenna. The curves in Fig. 1 and 2 show path loss as a func-
`tion of link range for a frequency near 1 GHz, and illustrate
`these dependencies for two different transmitter
`heights.
`These curves are typical of propagation of UHF, and the varia-
`tion of mean path loss as a function of frequency is typically
`much less than the variations
`due to terrain at a particular
`frequency. For example, the mean path loss from 700 MHz to
`2000 MHz varies about 8 dB, while path loss at a 20-km range
`is seen from the figure to exceed that of free space by 25 to
`and antenna heights. Further-
`80 dB depending on terrain
`more, the path loss in urban and suburban areas, where many
`area coverage packet radio net applications
`might occur, is
`more severe than that
`of most natural terrain. The
`curves
`shown reflect average values of
`path loss which apply to a
`link of a given length which is randomly selected without
`regard to user siting. Well sited radios will typically encounter
`less path loss than shown in the curves, while poorly sited
`
`
`
`
`
`PROCEEDINGS OF THE VOL. IEEE,
`
`
`
`
`
`66, NO. 11, NOVEMBER 1978
`
`radios will encounter larger path losses [ 351, [36]. These
`factors lead to large variations in achievable radio range among
`users and make RF connectivity difficult to predict in a large,
`mobile user community. The
`objective of packet radio net
`design is to overcome this difficulty without
`placing undue
`general, this
`restrictions on
`allowable user locations. In
`requires automated network management procedures capable
`of sensing the existing RF connectivity in
`real-time and
`instantly exploiting this connectivity for network control
`and packet routing.
`In addition to the wide variations in path loss, the ground-
`is subject tQ the effects of
`mobile, nonsited radio channel
`multipath propagation.
`When several
`differently delayed
`versions of the radio signal arrive at a receiver, constructive
`and destructive interference results. For stationary users, the
`effect of this phenomenon is that additional attenuation of
`the signal may be observed when the receiver is located at a
`point on the ground where the signal interference is destruc-
`tive. Nulls on the order of tens of decibels may be observed.
`When communicating users are in motion, or when a multipath
`component arises from a moving reflector, received signal
`strength fading is observed as a function of time. The rate of
`fading is proportional
`to the velocity of user motion. Move-
`ment by a mobile user of only a few meters can cause received
`signal strength to drop below the threshold of the receiver,
`thus effectively disabling the link. Radio connectivity
`to a
`mobile terminal may change dramatically even for small dis-
`placements, and for continuous motion,
`signal strength may
`fluctuate above and below the receiver threshold several times
`during the reception of a packet, causing several short bursts
`of errors in the data or even loss of synchronization altogether.
`We assume the modulation technique
`used by
`the packet
`is structured as a
`radio results in a transmitted
`signal which
`sequence of identifiible segments called symbols, where each
`symbol is one of a finite set of waveforms. In a simple binary
`modulation technique, one
`of two symbols is
`selected for
`each transmitted bit, depending on the value of the bit. In a
`spread spectrum system, the set of symbols may change with
`time, but a binary system would still choose each transmitted
`symbol from the set of two, which is in use at that particu-
`lar time. Typically, a receiver processes the arriving wavefom
`symbols one at a time, making a decision on each one as to
`which of the finite set has been received. The existence of
`multipath signal components affects the reliability with which
`symbol decisions can be made by causing symbol distortion
`and intersymbol interference. Intersymbol interference occurs
`when a symbol is overlapped by the delayed components of
`adjacent symbols. Such interference can lead to lower limits
`on symbol error probability which cannot be improved by
`increasing the signal to additive noise ratio on the radio link.
`When simple modulation techniques, such as phase-shift key-
`ing (PSK), are used, the symbol rate must be low enough that
`only a small portion of the symbol is overlapped by multipath
`