Enhancing GPS Accuracy and Security using DSRC
R-‐GPS (Robust GPS): Enhancing GPS Accuracy and Security using DSRC
VENKATESAN EKAMBARAM PhD Student Department of EECS University of California Berkeley
KANNAN RAMCHANDRAN Professor Department of EECS University of California Berkeley
1
TABLE OF CONTENTS EXECUTIVE SUMMARY .....................................................................................3 NEED FOR HIGH ACCURACY POSITIONING................................................4 A. FALLBACKS OF THE GPS SYSTEM ..................................................................................................................4 B. APPLICATIONS ENABLED BY HIGH ACCURACY POSITIONING ...................................................................5 B.1. Safety Applications ..................................................................................................................................... 5 B.2 Mobility Applications.................................................................................................................................. 6
NEED FOR SECURITY.........................................................................................7 EXISTING SOLUTIONS AND DRAWBACKS ..................................................8 PROPOSED TECHNOLOGY USING DSRC ......................................................9 DETECTING MALICIOUS USERS .............................................................................................................................12
CONCLUSION..................................................................................................... 13 BIBLIOGRAPHY................................................................................................ 14
2
Executive Summary
We propose a novel way of using the Dedicated Short Range Communication (DSRC) spectrum to enhance the accuracy and security of GPS. Traditionally the DSRC spectrum is designed to carry information such as the location estimates of the vehicles provided by GPS that are assumed to be accurate. However the accuracy of GPS is severely affected by environmental factors like multipath and adverse factors like intentional jamming and spoofing by malicious users. We ask the question of whether DSRC communication could in fact be exploited to enhance the position accuracy and security of GPS. Many safety and mobility applications as envisioned by the Federal Highway Authority (FHWA) for Intelligent Transportation System (ITS) applications, mandate sub-‐meter or higher accuracies that is not delivered by the GPS system. Department of Transportation (DoT)’s and other agencies are actively involved in exploring technologies such as NRTK, DGPS to provide a high accuracy location service for ITS applications. The cost of implementation and maintenance of these systems are expensive and further are not robust to multipath interference and malicious attacks. Our proposed solution aims to enhance the accuracy and security of GPS through collaboration between vehicles using DSRC. By exploiting the diversity in the DSRC measurements and using consistency checks, we aim to discard the bad measurements and obtain highly precise position estimates. The system can complement existing technologies such as NRTK to provide a robust, precise and secure location service for the benefit of ITS and other applications that mandate high accuracy positioning.
3
Need for High Accuracy Positioning In this section we will talk about the problems faced by GPS and the need for augmenting GPS for high positioning accuracy. We define high accuracy positioning to be sub-‐meter (decimeter) level accuracy. We will enlist the several applications related to transportation systems that would be benefitted by such a service.
A. Fallbacks of the GPS system
The Department of Defense operating the GPS constellation guarantees a location service accurate to 7 m for 97% of the time. Even though the level of accuracy suffices for most of the applications, there are various other safety and mobility related applications that demand sub-‐meter or lesser accuracy. The main problems faced by GPS are signal degradation and multipath interference that are particularly pronounced in harsh environments like urban canyon environments where the accuracies can be as bad as 50m or so. Figure 1 shows the example of a typical urban environment where the GPS signals get reflected off the surrounding buildings significantly degrading the positioning accuracy. The reflected signals introduce a bias in the estimates that lead to large errors in the position estimates. Solutions such as the AGPS, cellular triangulation improve the accuracy to some extent but are far from sub-‐meter accuracies. There are several upcoming GPS augmentation technologies such as NRTK, DGPS, HA-‐NDGPS etc that are aimed at providing sub-‐meter accuracies, which we will detail in the next chapter.
4
Figure 1. Multipath interference
B. Applications enabled by High Accuracy Positioning
There is a suite of ITS applications that benefit from a high accuracy positioning service. They can be broadly classified into safety and mobility applications. A comprehensive list of these applications is provided by the FHWA [1]. We will list a few of these applications that require sub-‐meter level or lane level accuracy that can be enabled by high accuracy positioning. B.1. Safety Applications These applications focus on reducing accidents and thereby fatalities and injuries.
5
a) Curve speed warning – These would aid the drivers in negotiating curves at appropriate speeds. There have been various studies that talk about the significant number of injuries and fatalities due to accidents that took place in curves. b) Forward Collision Warning – These alert a driver when a vehicle in the front brakes hard. c) Lane change warning – Warning if there is a vehicle occupying the blind spot. d) Intersection collision warning -‐ Warns a driver of a likely collision at upcoming intersections due their own speed or that of the other drivers. e) Left turn Assistant – Provides information of oncoming traffic when trying to take a left turn at an unprotected intersection. There are many other such safety applications that mandate sub-‐meter accuracies that are not possible with the present levels of accuracies guaranteed by GPS. B.2 Mobility Applications These applications aim at reducing delay, congestion, which can further have an impact on the environment. One of proposed applications is to have intelligent traffic controls. For example, if a vehicle were to know the signal phase timing at upcoming traffic signals, the speed could be optimally adjusted to reduce emissions. Further the traffic signal could be adjusted to provide priority to buses etc to encourage public transport. Another interesting application is to have free flow tolling and dynamic lane pricing systems, which can reduce toll plazas etc. All these applications require lane level accuracies that are not provided by GPS.
6
Need for Security
One of the major concerns regarding GPS is its vulnerability to spoofing and jamming. There is an excellent article [6] that talks about the dangers of GPS due to malicious users and unintentional interferers. The known signal structure of GPS renders it vulnerable to spoofing and jamming.
Figure 2: The dice is a small jammer (picture: [6]) Jamming is the intentional degradation of the GPS signal. Spoofing involves intentionally simulating a GPS signal in order to bias the position estimates of the GPS users. Jamming could sometimes be unintentional due to interference from strong transmissions in RF bands that are in the vicinity of the GPS transmission bands. Nevertheless addressing both of these issues is very important in order to have a robust system considering that the application involves the safety of the users. We will see how we could use DSRC and collaboration between vehicles to detect malicious users who can also intentionally bias their own location estimates to misguide the other vehicles.
7
Existing solutions and drawbacks There are various GPS augmentation technologies that have been proposed and implemented in order to enhance the accuracy. The most notable amongst these technologies include DGPS, N-‐RTK, HA-‐NDGPS and pseudolite based systems such as those proposed by Locata Corp[4,5]. DGPS, N-‐RTK, HA-‐NDGPS aim to improve the GPS positioning accuracy by providing location fixes calculated at a known reference station. These primarily correct for the ionospheric and tropospheric effects. Many of the state DoT’s are actively involved in deploying these systems in order to aid transportation applications. Examples of such deployments include the Ohio State RTK networks [2], Washington State Reference Network [3] etc. The pseudolite technology proposed by Locata involves transmitting GPS like signals from ground based reference stations and obtain the location estimate based on triangulation from three or more ground based reference stations whose locations are known. The proposed systems have certain drawbacks. These systems do not explicitly tackle the problem of multipath. Unless we have a highly dense deployment of reference stations, the problem of multipath would still exist. Secondly, the cost of establishing and maintaining these reference stations is quite expensive (around $500k to 1million per base stations, including the annual maintenance costs [2,3]). Thus, it would be highly beneficial to have an inexpensive system augment these reference stations in order to reduce the density of deployment and also tackle multipath. Our proposed system aims at achieving these objectives.
8
Proposed technology using DSRC GPS works by obtaining relative distance measurements from the target vehicle to three or more satellites and estimating the position using triangulation. The distance measurements are obtained by transmitting a known signal sequence over the allocated band and measuring the time delay in the signal reception, which is converted to a distance estimate. Locata works in a similar manner by obtaining distance measurements to ground based pseudolites having known locations. The high-‐level idea of the proposed technology is to convert each vehicle into a “virtual pseudolite” using the DSRC communication between the vehicles. This can be achieved by transmitting a known sequence of bits in the WAVE packet that would be exchanged between the vehicles. The sequence of bits can be processed to get a coarse estimate of the distance between the vehicles. The received measurements by each vehicle can then be efficiently processed to improve its location estimate by enforcing consistency amongst the different measurements. This is illustrated using an example in Figure 3. We have three cars with GPS and DSRC radios. The blue stars represent the predicted locations of the cars by GPS and the corresponding blue circles represent the position uncertainty of the GPS. By position uncertainty, we mean that the car could be located anywhere in the blue circle whose center would be the GPS location estimate. The error could be due to signal degradation, multipath etc. The black lines represent the estimate of the distance measurements between the cars obtained using the DSRC radios. The red circle imposes the consistency of the distance estimates.
9
Figure 3: Improving location accuracy using DSRC The overlap of the red and blue circles represents the new uncertainty in the vehicle locations, which is significantly smaller than the original uncertainty due to GPS. The red stars are the new estimates of the location that are significantly better than the GPS estimates. As one can see from this example, the position accuracy could be significantly improved using measurements from the DSRC radios. This is just a toy example to illustrate the benefits of using DSRC for positioning. One can come up with a more general and distributed signal processing algorithm in a larger setting where different cars get measurements with respect to their neighbors with whom they can talk to. The algorithms can be developed in a such a way that the processing happens only locally, i.e. each vehicle would only need to locally process
10
the measurements that it extracts from the DSRC packet received from its neighbors and estimate its location. A more detailed description of a sample algorithm that we propose can be found in [7] (a fairly theoretical paper), where we describe a decentralized algorithm to improve position estimates in a multipath environment. The effects of multipath are more elaborately modeled in the paper under some assumptions. Simulation results (in a restrictive setting) show that even if 90% of the readings are corrupted by multipath, one can obtain precise location estimates using collaboration. Practical issues such as time synchronization could be taken care of by using techniques such as time difference of arrival methods that require the DSRC radios to reflect back the packets that they receive. We could also have a centralized implementation wherein the road-‐side base stations or a centralized infrastructure could calculate the position estimates and feed them back to the vehicles. Our framework also provides a platform to integrate multiple technologies to improve the positioning. For example, WiFi, cellular, inertial navigation system, road side cameras, bluetooth etc could be processed locally by each vehicle and these could be exchanged between adjacent vehicles to improve their location accuracies. We could also integrate external sensors like multiple antennas that can operate over the DSRC band between vehicles to further enhance the position accuracies. Note that the proposed solution only aims to augments GPS and terrestrial ground based solutions for a more robust system and not replace it altogether.
11
Detecting malicious users
We can use the same idea of collaboration to detect malicious users. Lets consider the simple scenario of a malicious car on the road that reports its location incorrectly. The falsely provided position information could potentially lead to accidents, lane congestion etc. This could also reduce the confidence of the drivers on the warning systems thereby rendering them ineffective. However, by collaborating and using consistency checks, such malicious users can be detected and in some cases their positions could also be determined precisely.
Figure 4: Collaborative detection of malicious users using DSRC A simple example of detecting and estimating the positions of malicious users is shown in Figure 4. As before, consistency checks are imposed using the DSRC radio measurements between cars to detect if a car is reporting its true position. It is possible that the malicious car would also tamper with the DSRC signals. However, in this case, the malicious car would need to manipulate the measurements in such a
12
way that they all be consistent with a wrongly reported position. This would be hard to achieve when there are many neighbors with whom the consistency constraints would need to be imposed. In such a case, even though the position of the malicious car may not be determined, one can at least detect that there is a malicious vehicle whose messages can be ignored by the other vehicles. The idea could also be extended to the case where a user is trying to spoof or jam the GPS signals. The idea is similar to consensus based systems where redundancy and collaboration is used to detect abnormal users in the system.
Conclusion
Highly accurate positioning with sub-‐meter accuracy guarantee has a lot of potential applications both on the safety and mobility side for ITS. GPS does not guarantee the required level of accuracy for these applications and the existing solutions are expensive and do not address the problem of multipath errors. Further GPS is also prone to other security issues such as spoofing, and malicious users can affect the safety of other vehicles in the system. The proposed solution is a simple collaborative scheme that makes use of the DSRC band enabling the vehicles to behave as virtual pseudolites thereby increasing the accuracy and reliability of the system. The proposed technology in conjunction with augmented GPS systems such as NRTK, HA-‐NDGPS etc can be used to significantly improve the position estimates enabling a suite of applications that mandate sub-‐meter accuracies.
13
Bibliography
1. The CAMP Vehicle Safety Communications Consortium. Identify Intelligent Vehicle Safety Applications Enabled by DSRC. March 2005. 2. ODOT’s VRS RTK Network. Ohio Department of Transportation. http://www.dot.state.oh.us/Divisions/ProdMgt/Aerial/Pages/VRSRTK.aspx. 3. Washington State Reference Network. A Regional Cooperative of Real Time GPS Networks. http://wsrn2.org/. 4. J Barnes, C Rizos, J Wang, D Small, G Voigt and N Gambale, Locata: The positioning technology of the future, July 2003. 5. http://www.locatacorp.com/index2.html 6. http://mycoordinates.org/pdf/feb09.pdf 7. Venkatesan Ekambaram, Kannan Ramchandran, Distributed High Accuracy Peer-‐ to-‐Peer Localization in Mobile Multipath Environments, IEEE Globecom 2010, Miami Fl. http://www.eecs.berkeley.edu/~venkyne/venky2010p2ploc.pdf
14
