DSRC VANET Research Proposal
18-843: Mobile Computing Systems & Applications
Section A: Executive Summary
The purpose of this proposal is to bring to attention the physical layer issues, challenges, and research opportunities related to the use of the Dedicated Short Range Communications (DSRC) system in the various wireless communications applications of vehicular ad hoc networks (VANETs). The Antenna and Radio Communications Group in Carnegie Mellon University’s Department of Electrical and Computer Engineering currently has fully functional proof of concept VANET of five vehicles;1 however, very little has been done to evaluate the performance of the network and its usefulness in various applications.
There are three primary categories of useful applications for VANETs. First, VANETs can be used for vehicle safety applications, warning the cars within the network of close proximity, rapid braking, or other imminent danger. Second, they can be used for driver information applications such relaying slow traffic information or bottle-neck points. Finally, they can be used for entertainment applications and the exchange of other vehicle related software such as patches. These different applications have different throughput and quality of service requirements and the quality of the DSRC VANET communications channel must be evaluated to determine whether or not it is suitable for each application.
There are three primary areas which much be researched and documented before DSRC can be commercially deployed for any of these applications. First, there are several different environments in which a vehicle may traveland the network must meet the various minimum performance thresholds of each application in every situation. Second, there is a trade off between transmitting many small packets with a relatively large overhead vs. fewer large packets with a relatively small overhead and we must determine the packet size which optimizesnetwork performance. Finally, there are several possible rates at which the data can be transmitted and we must determine the data rate that yields maximum data throughput while meeting minimum network performance requirements.
We propose an extensive experimental research effort which utilizes and expands upon the existing proof of concept test-bed to collect large amounts of empirical data in various combinations of the three primary research areas. As we collect data, we will analyze it and determine the suitability of DSRC VANETs for each of the three primary potential applications.
Section B: Overview of DSRC
The Dedicated Short Range Communications (DSRC) system is a general purpose short to medium range RF communications link that supports both public safety and private operations in roadside to vehicle and vehicle to vehicle communication environments utilizing the IEEE 802.11p protocol. DSRC is meant to be a complement to cellular communications by providing very high data transfer rates in circumstances where minimizing latency in the communication link and isolating relatively small communication zones are important.2
The802.11a protocol, which is only months old, improves on the range and speed of transmission on the dedicated 5.85 – 5.925 GHz licensed band, promising a transmission range of about 1,000 feet and an average data rate of 6 Mbps.3
Section C: Problem Statement and Research Questions
With the increasing popularity of wireless communications technologies and their rapid expansion into most other technology sectors, the automotive industry is actively seeking ways to use wireless communications to create a competitive advantage in the marketplace. Researchers have identified three major categories of applications which will increase consumer demand for the automobiles with wireless technology – increased safety, driver information, and multi-media distribution. These different applications each have different throughput and quality of service requirements. On one hand, safety applications require very reliable communications links with very little latency, but do not need a very high bandwidth channel. On the other hand, the exchange of software, especially multimedia such as video, requires a rather large bandwidth, but is more tolerant of packet errors and latency.
The network performance requirements of the various applications are fairly well established, but unfortunately, the quality of the DSRC VANET communications channel and its suitability for each application is unknown. This raises the following research questions about DSRC VANETs:
- Can they meet minimum performance requirements in all possible driving environments?
- What is the packet size that optimizes link performance?
- What is the data rate that optimizes link performance?
- Are they capable of transmitting high priority messages that meet maximum allowable latency requirements?
- Are they capable of transmitting high priority messages that meet maximum allowable packet error rate requirements?
Section D: Research Goals
The primary goal of our research effort will be to determine whether or not DSRC VANETs are suitable for safety, driver information, and multi-media applications in vehicles.
The first major step toward achieving this goal will be to determine the effects of different driving environments on the signal propagation within the wireless communications channel. We will use the commonly accepted urban, suburban, and rural wireless communications environment categories. On one extreme, we must evaluate the large and small scale fading and other effects that are typical in multi-path dominated urban environments and on the other extreme, signal blockage and other issues that are typical in direct path dominated rural environments. Each environment also has unique traffic density features which must be evaluated if we wish to consider using multi-hop network routing. The overall performance and reliability of all wireless links are directly related to the environment in which the link is formed and we cannot establish the usefulness of DSRC VANETs if we cannot demonstrate that they meet minimum performance requirements in ALL environments.
The second major step toward achieving our primary goal will be to determine the effects of packet size on the overall performance of the wireless network. In general, there is a trade off between transmitting many small packets with a relatively large overhead vs. fewer large packets with a relatively small overhead. The network’s ability to tolerate and correct errors is directly related to the amount of data that is lost and the ease with which it can be retransmitted. If one large packet is lost, the effect is much more detrimental than it would be if one small packet is lost. On the other hand, the size of the packet header is independent of the size of the packet. As the ratio of packet header size to packet payload size increases, the efficiency of each packet transmission decreases. We must determine the point at which the packet is small enough that the system is still able to exceeded minimum performance thresholds given a dropped packet, but large enough that the ratio of packet payload to packet overhead is sufficiently large for adequate data flow.
The final major step toward achieving our primary goal will be to determine the effects of data rate on the overall performance of the wireless network. There are several possible rates at which the data can be transmitted; however, increases in data rate typically result in a decrease in the effective range of the transmissions within the network and the signal becomes much more vulnerable to dispersion, multi-path effects, fading, and other wireless communications channel effects. We must determine the data rate which optimizes data throughput in the network while meeting minimum performance requirements in sub-optimal environments.
After achieving each of these goals, will provide our research sponsors with the details of our findings and suggest a future course of action based on those findings.
Section E: Proposed Research Strategy
We propose an extensive empirical data collection effort which utilizes and expands upon the existing proof of concept test-bed to analyze the effects of various combinations of the factors in the three primary research areas. The execution of this research strategy is contingent on three major efforts.
The first major effort will be to improve CMU’s existing proof of concept VANET test-bed. At the moment, there are five vehicles or nodes in the network, each composed of a GPS unit, an IBM ThinkPad T23 laptop computer with an Atheros IEEE 802.11a mini-PCI card altered to emulate the DSRC Standard, a software package developed by our research group, and peripherals such as voice head sets and video cameras. We must upgrade the test-bed by phasing out the existing hardware and replacing it with DSRC hardware as it becomes commercially available.
In addition, we need to develop or otherwise acquire a means of measuring precise location in urban environments. Recent data collection efforts demonstrate that GPS technology is not a sufficient means of making reliable location measurements in urban environments due to satellite signal blockages, multi-path signal reflections, and other factors which limit its ability to maintain a consistent communications link with enough satellites.
The second major effort will be to conduct the numerous data collection drives required to collect enough data to thoroughly evaluate each possible combination of driving environment, packet size, and data rate. Our initial emphasis will be on data rates of 3, 6, and 24 Mbps and packet sizes of 50, 100, and 250 bytes in each of the three environments. We estimate that twenty hours of driving in each scenario ought to be sufficient to make a well-substantiated determination of network performance. The actual data collection strategy is fairly simple – begin with one combination (i.e. a data rate of 3 Mbps and a packet size of 50 bytes in a rural environment) rotate through the various combinations until all have been accounted for.
The third and final major effort will be to analyze the data, evaluate the network performance in each of the scenarios, and make a final determination of whether or not DSRC VANETs are suitable for safety, driver information, and multi-media applications in vehicles. We have created extensive MatLab analysis tools for evaluating transmission link performance using such metrics as packet error rate. We must continue to refine these tools and further develop their capabilities as we explore new ways to evaluate network performance. The final determination will be based upon a comparison of the minimum performance requirements of each of the possible applications and the minimum performance levels of the empirical network data for ALL possible scenarios.
Section F: Research Metrics
In order to achieve success in this research effort, we must define the minimum application requirements and establish some way of measuring the network performance.
The primary limiting factors for safety applications are link dependability and latency. If the link is not dependable, then they cannot be entrusted with the safety of the driver and passengers of the vehicle. If the safety message takes too long to reach the driver, either to accident will occur or the driver will observe the danger and take preventative actions on his own and the safety application will accomplish nothing. Current system designs call for redundancy in safety messages, so a link is considered to be dependable if it has an overall bit error rate of 0.005 or less. The average human reaction time is estimated to be about 500 msec, so total system latencies of less than 250 msec are considered small enough that the application will benefit the driver.
The primary limiting factor for multi-media applications is transmission bandwidth. The application with the greatest bandwidth demands at this point is video conferencing, and depending on the type of application used, video conferencing requires a bandwidth of at least 32 kbps. Voice applications and driver information applications fit somewhere between the two extremes.
The MatLab analysis tools that our research group has developed are currently capable of measuring packet error rates as a function of several variables including distance, received signal strength, and absolute and relative speed. As each data set is collected, the user is able to set the maximum data rate and packet size. We must further develop our analysis software to record actual data rate, and once we have done so, we will be able to measure empirical values of each of the metrics used to define network performance as described above.
Section G: Required Resources
The successful completion of this research project will require hardware, software, and manpower. We will require a minimum of five vehicles, complete with VANET communications test kits, to evaluate the network. Much of the software that will be required can be developed by the research group; however, we anticipate some expense in purchasing additional software and licenses. In addition, we will require at least one university faculty advisor, two PhD program students, two Master of Science program students, and four technical staff to maintain and drive the vehicles.
Section H: Future Research Opportunities
The focus of this research proposal has been on the physical layer issues, challenges, and research opportunities related to the use of the DSRC system in the various applications of VANETs with an emphasis on the wireless communications channel. If we are able to establish the fact that DSRC VANETs are suitable for the applications that we foresee, there will be many opportunities for additional research work! At the moment there are several ad hoc networking protocols such as AODV and DSR, however, none of them are optimized for dynamic mobile environments. There is a great need to develop a mobile ad hoc networking protocol that better addresses the needs of VANETs and expands the capability, performance, and efficiency of the network. Finally, the expansion of wireless capabilities to the automotive world creates a great opportunity to develop new applications that go above and beyond those of the potential applications which motivate this research work.
Section I: Expected Impact of Research
We are certain that our research effort will make significant contributions to the ongoing and widespread effort to expand wireless communications to vehicular applications. DSRC is actively supported by the U.S. government and the Intelligent Transportation Systems (ITS) organization. The research at CMU is aggressively supported by General Motors and several other automobile manufacturers are also pursuing this technology and the many application opportunities it creates. Our research will have a direct role in determining the whether or not the DSRC system is suitable for the future development and commercialization of VANETs, and if so, for which applications it is most suitable.
Section J: References
- Mangharam, Rahul, J. Meyers, et al., A Multi-Hop Mobile Networking Test Bed for Telematics. SAE International, 2004.
Page 1 of 5