I wrote this term paper for ECE 463, Communications II, in the Spring of 1991.


Packet radio (PR) networks are used to transmit data over long distances. While their efficiency is adequate for present applications, the use of spread spectrum coding has the potential of greatly expanding the capacity of PR networks. In addition to increased capacity, spread spectrum can provide message privacy for those applications that need it. Work on spread spectrum in PR networks is in the experimental, prototype, and/or classified stage at this point. Because the technology involved is so interesting, this report will examine the nature of these systems even though they are not yet in use.


Packet radio is used by a network of users to communicate computer data back and forth. It began in 1970 with the ALOHA network at the University of Hawaii. They used PR to communicate with the different islands of Hawaii. Since 1980, when the FCC made a temporary allocation of frequency space to this purpose, an amateur packet radio network has sprouted up in this country, today with thousands of users. PR is essentially packet switching taken to the airwaves [1].

Packet switching is a transmission process where binary message data is grouped into finite-length packets, given address information, and sent out on a computer network to seek the destination of its address. Several users use the same data channel for efficiency. When a pair of messages from separate users appear on the network simultaneously, the packets collide and the information from both is lost. Then both packets have to be resent.

A normal packet network is structured into clusters. A group of users locally work form a cluster and communicate over the same channel to each other. For communication to users outside of the local group, one of the local computers must act as a clusterhead. The network of clusterheads uses a separate communications channel for its communications. If the destination user belongs to cluster far far away from the source, clusterheads between the source and destination will receive the packets and retransmit them to the next cluster until the destination cluster is finally reached.

With PR's increasing popularity, spread spectrum coding is being looked at as a long term solution to providing enough capacity for the ever increasing amount of message traffic. With the limited funding available to universities and amateur radio, experiments are so far rather limited. Another group looking to apply spread spectrum coding in packet radio has a considerable amount of money to apply to research: the US military.

Under development for the US Navy is the High Frequency (HF) Intra Task Force (ITF) Network [2]. The object is to provide fast, secure, jam-resistant data communications between the ships of a naval taskforce. A PR network designed to do this must be very flexible and robust. This system is now examined.


The requirements of this system are challenging. The HF frequency range to be used is 3-30 MHz. The HF ITF network should connect 2-100 users with voice and data communications at a rate somewhat less than 10 kbits/s/link. A member of the task force should be able to communicate with other members of the task force and also with members outside the task force. The connectivities of the network must be dynamic as the position of nodes changes as the ships move around or disengage entirely from the task force. It has to deal with the threat of hostile jamming of its communications and the possibility that some of the ships forming the network can be destroyed by hostile action. The communications must be secure, not able to be deciphered by the enemy and hopefully not giving away the task force's location either. And both point-to-point and broadcast communications must be allowed, with various message precedent levels supported.


To provide a flexible network, able to cope with jamming and the dynamics of the members, control of the network must not be fixed and centralized. Control will be distributed among the various nodes, with each node having the same instructions. In forming the network, the nodes will query and acknowledge their neighbors until they can agree on the formation of clusters based around a few of the nodes.

Spread spectrum will be used because of its anti-jam capability and its ability to allow simultaneous transmission over the same channel. With spread spectrum, a signal is spread out over a much larger bandwidth than that which is required for normal transmission. The result is that the signal strength over a narrow range of frequencies is very low. Two spread spectrum techniques are considered here, frequency hopping and direct sequence.

In frequency hopping (FH), a transmitter broadcasts a strong signal over a narrow frequency range, but rapidly shifts the frequency range at which it is transmitting. The net effect is that the transmission is spread over a wide range of frequencies when a sufficient length of time is considered. In order to pick up the transmission, the receiver must know ahead of time the sequence of frequencies that will be used by the transmitter.

When more than one FH transmitter is using the same frequency range, there is the chance of error. Collision errors occur when two or more transmitters transmit over the same frequency range at the same time, wiping about the information from both signals. For a small number of users over a large enough frequency range, the chances of landing on the same frequency at the same time are very small. Error correction coding can usually take out errors due to collisions.

Direct sequence (DS) spread spectrum uses a very fast binary bit stream to shift the phase carrier frequency. The distribution is supposed to appear to be random, that is, the binary bits are called pseudonoise (PN). Here as well, the receiver must know exactly the sequence of PN used by the transmitter in order to pick up the message. The receiver interprets the message by correlating the strength of the signal received at various frequencies with the PN coefficients used when the signal was transmitted.

Code division multiple access (CDMA) is the concept of many users being able to share the same frequency range simultaneously. Because different transmitter-receiver pairs correlate their signals with different PN codes (DS) or shift frequencies differently (FH), more than one system can communicate at the same time over the same channel. Using CDMA in PR networks eliminates the problem of two packets colliding and both being destroyed.

Important for HF ITF Networks is the ability of spread spectrum to resist jamming. With FH, the jammer can not follow the system fast enough as it switches from one frequency to another. With DS, although the jammer can try to spread its power out and jam all frequencies at the same time, it takes a very powerful jammer, otherwise correlation of the PN codes with the broad signal brings back out the signal transmitted even in the presence of significant noise or jamming [3].


The HF ITF Network will use the Linked Cluster Algorithm (LCA), a distributed algorithm, to establish the connectivity of the network. Each node is assigned an ID number. To start off, a node transmits its ID number out to other nodes, and acknowledges back to the other nodes the ID numbers it has received from them. If a node has the lowest ID number of all nodes that it can hear, it declares itself to be a clusterhead. Nodes that are within communications range of two or more other clusterheads declare themselves as gateways and will carry traffic back and forth between the clusterheads.

The signals in the HF band travel mostly along the ground and have a fairly short range; the range increasing with frequency. In order to fully utilize the variation of range with frequency and to provide more robustness, the HF band is broken up into a few frequency divisions. Separate but interconnecting networks are based around each frequency division, with the highest frequency networks having the longest range. These separate networks can each be utilized to transmit data and if one fails the others can back it up and help it to reestablish contact.

Time division multiple access (TDMA) is used for the bare-bones network of the clusterheads in their management of the network and in their control of the routing within their cluster. Frequency hopping with its anti-jam and multiple-access ability is used to handle the bulk of data transmissions. The basic coding method is noncoherent M-ary FSK with one M-ary symbol per hop and appropriate error coding. Frequency hopping was chosen over direct sequence spread spectrum because of DS high sensitivity to received power levels. FH was more robust . The results at this stage seem promising


It can be expected that the use of spread spectrum will become quite common in packet radio networks. In general, there is a lot of potential for spread spectrum technology.

Personal communications networks are being researched, where spread spectrum will become competitive with cellular telephone technology. The FCC has a lot of say on what happens next with spread spectrum because it requires a lot of bandwidth and the assigned radio spectrum is mostly full.



[1] The ARRL Handbook for Radio Amateurs, 68th ed., The American Radio Relay League, Inc., 1990, p. 19-24.

[2] A. Ephremides, J. E. Wieselthier, and D. J. Baker, "A design concept for reliable mobile radio networks with frequency hopping signaling", Proceeding of the IEEE, vol. 75, pp. 56-73, Jan. 1987.

[3] D. L. Schilling, L. B. Milstein, R. L. Pickholtz, M. Kullback, and F. Miller, "Spread spectrum for commercial communications", IEEE Communications Magazine, vol. 29, pp. 66-79, April 1991.

Writing Samples 

Scott's HomePage