Car to Car Communication for AVCS
Contributed by Bret Foreman
Automated control of passenger vehicles is one promising method of increasing traffic density and efficiency on existing highways. Most schemes will require extensive use of wireless communication for coordination of maneuvers, stable control, safety, traffic advisories, navigation, and fault control. Wireless communication between pairs of closely spaced vehicles, maneuver (communication among groups of vehicles), and advisory (communication between vehicles and a central, stationary database) functions are needed. In this context, the control function is to maintain stability and good ride quality among the vehicles in a group; the maneuver function is for a vehicle to join or leave a group; and the advisory function is to inform the groups about conditions ahead.
Transmission aspects
In a mobile environment, the issues of modulation, directionality, and addressing are all closely related. Broadly speaking, the communication channels available in a mobile environment may be line-of-sight or broadcast.
Addressing
Addressing may be local or global. Modulation may be used to create multiple channels in time, frequency, spreading codes, etc. The number of combinations is large. Of course, certain combinations are optimal and we will discuss which and why below. First, a summary of the properties of the different options : A line-of-sight system may use short addresses consisting of lane number and platoon sequence number (the first vehicle in a platoon is number 1, the second is number 2, etc.).
This provides enough addressing to prevent ambiguity at the local level due to the physical limit on the number of vehicles that can "see" the transmitter. An 8 bit address is sufficient to encode lane numbers from 1 to 8 and platoon sequence numbers from 1 to 32. This is a good example of local addressing. Broadcast systems cannot take advantage of such fortunate physical configurations and must use global addressing. With global addressing, each vehicle has a globally unique address, much like a license plate number.
First Contact and Communication
Another important consideration is the problem of "first contact". First contact occurs when one vehicle or platoon approaches another in the same lane. The initial problem is one of mapping a sensor target (usually a radar reflection) to an address that can be used to hail the approaching target. In a line-of-sight system the hailer can simply cycle through the 32 possible addresses that may occupy that lane until a response is returned. In the global/broadcast case, the only solution that this author has seen is to modulate a radar reflector on each vehicle to provide a unique radar signature.
This is a very complex technique and requires close coordination between the communication system and the sensor system. Line-of-sight systems are usually optical but there is no technical reason why micr waves could not be used.
Commercial micr wave systems are mostly too narrow-beam, too long-range, and too costly for AVCS needs but a custom system is possible. Falling somewhere in between line-of-sight and broadcast are radio systems with dual-directional antennas (also called dual-diversity antennas). These systems are not precise enough to use local addressing or to solve the first contact problem but they do, at least, provide an indication of whether a target is in front or in back. The added gain of a directional antenna can help extend the range of the system. More importantly, the ability to switch antennas can be used
to mitigate fading. (See also
car-to-car propagation) Many of the commercially available radio systems feature dual-directional antennas for this reason.
The problem of fading leads naturally to a discussion of modulation. In the closely spaced platoon propagation environment the dominant fading mechanism is multipath interference. [ 9 ] This is only an issue in radio systems where the wavelength of the carrier is greater than a few millimeters. That includes all the commercial digital radios operating in the 900 MHz and 2.4 GHz bands. These types of radios are very attractive because of their widespread commercial use, their low cost, and their relative freedom from regulation. That light regulation is a result of the spread spectrum modulation technologies used in these bands.
Unfortunately, some types of spread spectrum modulation are very adversely affected by the signal strength fluctuations associated with multipath interference. We have studied two spread spectrum modulation technologies, the first being direct sequence and the second being frequency hopping FM. [ 10 ]
Direct sequence
Direct sequence modulation is performed by multiplying the serial digital message at, perhaps, 2 Mbits/second by a "spreading code" at, perhaps 64 Mbits/second using an exclusive OR logic function. The resulting digital data stream is then used to AM modulate a 2.4 GHz carrier. After standard AM demodulation, the original digital message may be recovered by performing another exclusive OR with the same spreading code. Of course, the receiver's spreading code must be synchronized with the incoming signal and this requires a "preamble" or known bit pattern at the beginning of each packet. These long (over 100 bit) preambles are a major source of overhead in AVCS control applications where packets are short. Though the time added by the preamble (50 microsecond) is
negligible when compared to the time between packets (20 mS), the preamble more than doubles the size of the packet and thus increases the sensitivity to bit errors significantly. [ 11 ] Since the underlying AM modulation is very sensitive to the signal power fluctuations common in a mobile environment, bit error rates are high. Lastly, our own experimentation has shown that direct sequence spread spectrum radios suffer from extreme degradation in range and error rate if the stations are moving with respect to each other at more than a few meters per second. This behavior has been confirmed (though not completely explained) by the radio manufacturers themselves. According to the theory presented by one radio manufacturer, this effect scales linearly with relative velocity and packet size. They claim that small packets can still be sent, even at high relative velocities. We are currently working to prove or disprove this theory. For these reasons direct sequence spread spectrum is not optimal for AVCS applications, though it can be made to work at some cost in efficiency.
Frequency Hopping
Frequency hopping spread spectrum uses a tunable FM transmitter and receiver with about 75 detectable carrier frequencies. The radios change or "hop" frequencies 75 times a second.
All radios on the same "channel" use a common hopping sequence. FM modulation has a much broader bandwidth than AM to start with and the added frequency hopping increases this further. In the most sophisticated systems the hopping sequence can change adaptively to avoid frequencies that exhibit excessive fading. The most attractive feature of these radios is that FM modulation is very resistant to fading in a mobile environment. Another advantage for AVCS control systems is that long preambles are not necessary for synchronization. Bit synchronization can typically be accomplished in less than 20 uS or 20 bits at a typical 1 Mbit/second data rate. This means that short packets are not excessively lengthened by the preamble.
Redundancy of Control-Communications
- Since the quality of control depends on the quality of communication, it is important to provide some redundancy in the control-communication system. One way to do this is to enable the communication routing software to re-route control packets to the maneuver-communication system in the event of a control-communication fault. This requires the maneuver-communication system to have some extra capacity to handle the contingent traffic. A natural question to ask is, "If the maneuver communications system needs to have the capacity to handle the control information too, why have the control communication system at all?" The answer is that the broadcast maneuver-communication system actually shares bandwidth with all the platoons in the neighborhood. Each line-of-sight control channel that we replace with maneuver bandwidth takes capacity away from all these neighboring platoons. Also, addressing in the maneuver channel is global, slowing down communication because of the added overhead and adding to the bandwidth required. For these two reasons the maneuver channel should be used sparingly - only for maneuvers and emergency control.
Solution Options
Control Options
- In the PATH program, we use an infrared (IR) system for line-of-sight communication. Though an RF line-of-sight system might give superior performance in bad weather, the IR system is simpler, cheaper, and provides better control of beam coverage. The Infrared Line-of-Sight channel - The infrared communication system is optimized for mobile applications. This system uses IR radiation in the 830 nm band modulated with on-off-keying. The clock is encoded in the data. It has a useful range of about 30 meters and a beam spread that can be very precisely controlled. [ 12 ] It uses an adaptive data rate controlled by a protocol similar to that used in variable-rate modems. [ 13 ] The data rate varies inversely with distance from 19.2 Kbits/Sec to 1.2 Mbits/Sec, depending on the bit error-rate. Since the clock is encoded in the data, the maximum channel bandwidth is about 3 MHz.
Medium Modulation
Infrared On-Off Keying w/ Clock Encoding
Desirable Characteristics
- Simple addressing
- No FCC restrictions
- Good for first contact
- Well controlled beam coverage
- Faster data rates at close range
Undesirable Characteristics
- Beam is attenuated by rain, snow, dirt, etc.
Radio Direct Sequence Spread Spectrum
Desirable Characteristics
- Complete commercial systems available
- Liberal FCC rules
Undesirable Characteristics
- Fast fading problems
- Limits on relative velocity
- Not optimized for AVCS applications
Radio Frequency Hopping Spread Spectrum
Undesirable Characteristics
- Easily optimized OEM system
- Liberal FCC rules
- Customizable modulation
Undesirable Characteristics
- Some custom hardware required
- Less support from Manufacturer
Data Link
The Datalink layer packet structure is HDLC.[ 14 ][ 15 ] This structure provides 8 bits of addressing and a 16 bit error checking code. The maximum packet size supported is 64 Kbytes but in practice packet sizes larger than 1 Kbyte are rarely used. There is an additional overhead of 5 bytes associated with synchronization, start/end flags, and control. We use control packets of about 16 bytes. The system also has the capability to carry other information interleaved with the control packets. [ 16 ] These are called data packets in reference 16 and may be up to 100 bytes.
Maneuver Options
- Vehicles entering and leaving platoons must coordinate their actions. [ 4 ] This requires a certain amount of low bandwidth communication. For obvious reasons, this communication cannot be line-of-sight. Because of the low bandwidth and the omnidirectional nature of the channel, broadcast is the method of choice.
Advisory and Navigation Options
- A good solution should have at least the following properties: It should be long range. It should have enough bandwidth for traffic advisories. It should have expandable capacity to deal with future enhancements and added features.
Experiments with commercially available radios
The PROXIM radio channel
- The current vehicle control system uses radio for transfer of control packets. The radios are built by PROXIM. [ 17 ] They operate on a single direct sequence spread spectrum channel which is time-domain multiplexed among the vehicles in a token-bus protocol. [ 18 ][ 19 ] The data rate is 256 Kbits/Sec. The long "turn around" time of these radios (time from transmit to receive mode) slows the token-bus protocol down about 8 mS per vehicle. The "Hand Off" time consists primarily of the radio turn-around time and some processing overhead. This long turn-around time is due to the low bit rate and the long preamble plus some delay in the serial control of the receiver. While these may be fine radios for some applications, they are not optimized for a token-bus or for mobile environments. We see a much reduced range in a high-speed mobile environment using this direct sequence modulation (as would be predicted from the discussion above). The manufacturer's nominal range is 500 meters but we measure about 100 meters at freeway speeds.
The Pulse Radio channel
- The Pulse radio system uses frequency hopping spread spectrum modulation. [ 20 ] It has a nominal range of 500 meters and a bit rate of 1 Mbit/Sec. These radios are still under development in the PATH program so we do not have any performance data at this time.
Because of the more optimal modulation, we expect the range at freeway speeds to be similar to the stationary range (nominally 500 meters). The manufacturer specifies a turn around time of 20 uS, far faster than the current system. As well as being used for maneuver-communication, this radio is efficient enough to be used in a very fast token-bus that could support control information. This could provide redundancy for a line-of-sight control-communication system.
The WaveLAN radio channel
- The WaveLAN radio system uses direct sequence spread spectrum modulation. [ 21 ] This modulation technique limits the length of the packets that may be used at freeway speeds. However, the limit is on packet transmission time, not bit length. Since the WaveLAN radio has a data rate of 2 Mbits/Sec the maximum packet length is 100 bytes at a relative velocity of 30 meters/second. That is large enough for both control and maneuver. These radios are still under development in the PATH program so we do not have any performance data at this time. We expect that there will be some degradation from the manufacturer's specified range (500 meters) at freeway speeds due to the type of modulation used. A nice feature of the WaveLAN radio is that it is a commercial product sold in the form of a turn-key wireless networking system. That means that the price is competitive and the system hardware is simple to set up. It is still necessary to write custom driver software to allow the radios to work in a platoon environment but at least the hardware is complete.
References
- 1 ) Highway Capacity Manual, Special Report 209, Transportation Research Board, National Research Council, Washington, DC, (May, 1992).
- 2 ) "Highway Infrastructure: Quality Improvements Would Safeguard Billions of Dollars Already Invested", pp 50, General Accounting Office, Washington, DC, (1994).
- 3 ) P. Varaiya and S. Shladover, "Sketch of an IVHS Systems Architecture", Institute of Transportation Studies, University of California at Berkeley, PATH Research Report UCB-ITS PRR-91-3, (February 1991).
- 4 ) A. Hsu and F. Eskafi and S. Sacks and P. Varaiya, "Protocol design for an automated highway system", Discrete Event Dynamical Systems, 1 (2) pp 183-206, (1993).
- 5 ) Hedrick, J. K., Swaroop, D. "Dynamic Coupling in Vehicles Under Automatic Control", Proceedings of the 13th IAVSD Symposium, Chengdu, China (August, 1993).
- 6 ) S. L. Streisand and J. Walrand, "A Communications Architecture for IVHS", Institute of Transportation Studies, University of California at Berkeley, PATH Research Report UCB-ITS-PRR-92-10, ISSN 1055-1425, (December 1992).
- 7 ) S. Low and P. Varaiya, "A New Approach to service provisioning in ATM Networks", IEEE/ACM Transactions on Networking, 1 (5) pp 547-553, (October, 1993).
- 8 ) I. Hsu and J. Walrand, "Communication Requirements and Network Design for IVHS", Institute of Transportation Studies, University of California at Berkeley, PATH Technical Report Number 93-18, (August, 1993).
- 9 ) J.S. Davis II and J.P.M.G. Linnartz, "Vehicle to Vehicle RF Propagation Measurements,", 28th Asilomar Conf. on Signals, Systems, and Computers, Pacific Grove, California, pp 470-474, (November 1994).
- 10 ) E. Lee and D. Messerschmitt, Digital Communications, pp 445-450, Kluwer Academic Publishers, (1988).
- 11 ) Jean Walrand, Communication Networks: A First Course, pp 122, Aksen Associates Incorporated Publishers (1991).
- 12 ) B. Foreman and B. Schein and S. Streisand, "An Infra red Inter-Vehicle Communication System", submitted for publication in The Journal of Vehicular Technology, (1994), e-mail to foreman@eecs.berkeley.edu for details.
- 13 ) Eskafi, Nassiri-Toussi and Liu, "Adaptive Baud-Rate Protocol For Wireless Communication", 1994, in preparation, e-mail to farokh@eecs.berkeley.edu for details.
- 14 ) J. Walrand, Communication Networks: A First Course, pp 46-54, Aksen Associates Incorporated Publishers (1991).
- 15 ) International Organization for Standardization. High level data link control (HDLC) procedures - Frame structure. ISO 3309, (1991).
- 16 ) H. Huang and B. Foreman, "Infrared Datalink Layer Documentation", Institute of Transportation Studies, University of California at Berkeley, PATH Technical Note 95-2 (January, 1995). 7
- 17 ) Proxim, Inc., 295 North Bernardo Ave., Mountain View, CA 94043, Phone 415-960-1630.
- 18 ) W. Li, "Design and Implementation of Digital Radio Communications Link for Platoon Control Experiments", Institute of Transportation Studies, University of California at Berkeley, PATH Research Report UCB-ITS-PRR-95-2, (January, 1995).
- 19 ) Jean Walrand, Communication Networks: A First Course, pp 45, Aksen Associates Incorporated Publishers (1991).
