Requirements for the Physical Layer of the NOTICE System for Vehicular Communications

Transcription

1 Requirements for the Physical Layer of the NOTCE System for Vehicular Communications Dimitrie C. Popescu and Dusadee Treeumnuk Department of Electrical and Computer Engineering Old Dominion Uniersity Norfolk, VA Stephan Olariu Department of Computer Science Old Dominion Uniersity Norfolk, VA Abstract n this paper we inestigate physical layer requirements for the NOTCE system proposed for notification of traffic incidents. The system uses short-range wireless communication between ehicles and sensor belts embedded in the roadway which impose constraints on time aailable for connection and data exchange, and we study the probabilities of establishing the wireless link and of successfully exchanging information between a sensor belt and a ehicle passing oer it. We derie analytical expressions for these probabilities as functions of seeral parameters such as the time aailable for handshaking/information exchange, aerage speed of the ehicle, data rate and amount of information to be exchanged between ehicle and belt, and we ealuate their alues for specific parameters corresponding to practical scenarios. Our numerical results indicate that inexpensie short-range wireless systems such as ZigBee radios are good candidates for the physical layer of the NOTCE system.. NTRODUCTON According to estimates by the United States Department of Transportation (US-DOT) congestion eents on major roads and highways due to traffic incidents result in huge losses in worker productiity and billions of gallons of fuels wasted []. The US-DOT also notes [] that oer half of the congestion eents are caused by highway incidents rather than by rush hour traffic in densely populated areas. With sufficient adance notification of traffic congestion driers could make educated decisions about taking alternate, less congested routes, which would improe the oerall traffic fluency and reduce the seerity of congestions. Current technology used for traffic monitoring and incident detection based on nductie Loop Detectors (LDs) is expensie and not ery accurate and reliable, and the ntelligent Transportation Systems (TS) community is looking for less expensie and more reliable methods for traffic monitoring and incident detection. Vehicular Ad-hoc Networks (VANETs) employing latest wireless sensor and networking technologies [2] hae receied increasing attention in the TS community lately. They employ a combination of Vehicle-to-Vehicle (V2V) and Vehicle-to-nfrastructure (V2) wireless communication and are enisioned to integrate the driing experience into a ubiquitous and perasie network that will enable noel solutions for traffic monitoring and incident detection [3], [4]. Recently, a secure and priacy aware architecture dubbed NOTCE was proposed for implementing a new system for notification of traffic incidents [5]. NOTCE is a V2 system whose architecture enables concrete security and priacy mechanisms by employing short-range secure radio communication operating at low power which preents entities that are not actual traffic participants to gain access to the system and does not disclose any identifying information for ehicles in traffic. A related approach based on using a different type of infrastructure was proposed in [6]. Howeer, the use of short range wireless communications imposes strict constraints for the NOTCE system as ehicles on the road hae a limited time to establish connection with the roadway infrastructure and to successfully exchange traffic related information. We note that the actual speed of a gien ehicle determines the time that the ehicle has aailable for communication and information exchange with the road infrastructure and this should be closely correlated with the specifications of the wireless technology employed. n this paper we introduce formal requirements for the physical layer of the NOTCE system and inestigate the probability of establishing a wireless link as well as the probability of successful data transfer between ehicles and the roadway infrastructure. The paper is organized as follows: in Section we gie a brief reiew of the NOTCE system followed by an analysis of the wireless communication process between ehicles and the roadway infrastructure in Section. We present numerical results in Section V and final remarks and conclusions in Section V.. A BREF REVEW OF NOTCE According to [5] the NOTCE architecture assumes that ehicles are equipped with a tamper-resistant Eent Data Recorder (EDR) designed to store basic information about the ehicle s moement such as lane changes, speed, acceleration, position, as well as any additional information that may be input by the drier related to actual road conditions. The EDR works in conjunction with two short-range radios installed in the ehicle that operate at low power to communicate with wireless sensor belts embedded in the roadway at some gien distances designed to collect the traffic information and proide drier notification in case of incidents. This is different than the approach in [6] where magnetic sensors are deployed 3

2 along both sides of the road for information collection and dissemination. Handshaking Time (th) Connection Setup Data Exchanging Time (td) Data Exchange # radio for handshaking #2 radio for data exchange HELLO sent by belt ACK receied by belt Complete V2B Communication r r 2 r 2 Communication Range of Transceier # Communication Range of Transceier #2 Fig.. Vehicle equipped with EDR and two short-range radios for the NOTCE system. HELLO sent by belt ACK receied by belt r2 r r2 Sensor belt Data exchange between ehicle and belt Fig. 2. Sketch of communication between a ehicle and a sensor belt in NOTCE. The two radios inside the ehicle are placed as shown in Fig. with one radio mounted close to the front axle of the ehicle and the second one installed at the rear of the ehicle. The corresponding communication ranges of the two radios are also shown in Fig. and 2, and we assume that r and r 2 shown in these figures are the same for all ehicles. We note that this is a reasonable assumption if one considers that similar wireless technology implied by some gien standard is used by all ehicles in the NOTCE system. The communication process between a belt and a ehicle is sketched in Fig. 2 and is as follows: when the wireless sensor in the belt determines that a ehicle has arried (using for example pressure sensing actiated by the front wheels of the ehicle) a radio transceier on the belt will attempt to establish communication with # transceier in the ehicle by sending a short range HELLO message with low transmit power. While the sensor belt is within the range r of transceier # in the ehicle this has limited time t h aailable to respond with an acknowledgment message ACK to the HELLO message from the belt in order to complete the handshaking stage of the communication between ehicle and sensor belt. f the belt does not get the acknowledgment from the ehicle it will not attempt further communication with that ehicle. Upon successful handshaking between belt and ehicle a wireless radio link is established with #2 transceier in the ehicle and information/data will be exchanged securely by Fig. 3. Sketch of the timing diagram for car-to-belt communication in NOTCE. using symmetric key encryption technique as discussed in [5]. The transmission power aailable for the information/data exchange will be higher than that used in handshaking. Consequently, as shown in Fig. 2, the range 2r 2 of transceier #2 in the ehicle is larger than r and, assuming constant ehicle speed, a longer time interal t d is aailable for data exchange than the time t h aailable for handshaking as shown in the timing diagram of Fig. 3. Unlike other systems proposed for notification of traffic incidents where traffic participants alert each other directly of incidents [7], in the NOTCE system a gien sensor belt first confirms that a traffic incident has actually occurred by aggregating information successfully receied from multiple ehicles and then sends alert messages to traffic participants. Belts in the same driing direction may communicate with each other indirectly through passing ehicles that carry information from a gien belt to the next belt: a passing ehicle uploads information receied from preious belt to the current belt and also downloads information from the current for the next belt. Sub-belts on opposite sides of the road are linked together and communicate directly when a gien belt needs to send notification about incidents to other belts in the backward direction. The incident detection time, which is the time needed by a sensor belt to determine that a traffic incident has occurred, is influenced by arious parameters such as the spatial density of the ehicles in traffic, the spacing of the sensor belts, how conseratie the belt inference mechanism is, and the number of ehicles that are able to exchange information with the sensor belts [5]. We note that an acceptable alue for the incident detection time of around minute can be achieed een when only a fraction η = of the ehicles that pass oer a gien belt exchange traffic-related information with the belt [8]. n the following section we present an analysis of the requirements associated with the actual communication between a passing ehicle and a sensor belt, and we derie an expression for the probability of successful information exchange between ehicles and sensor belts.. TMNG REQUREMENTS FOR VEHCLE-TO-BELT COMMUNCATON To inestigate the timing requirements of the ehicle-tobelt communication in the NOTCE system we first conert the range parameters r and r 2 of the two radios mounted in the ehicle to aailable time alues as follows: r = T ah k () 2r 2 = T ad k 2 (2) 4

3 where T ah is the maximum time aailable for handshaking, T ad is the maximum time aailable for data exchange, and k, k 2, are conersion factors with units of measure of speed. We note that the alues of T ah and T ad depend on the wireless technology and standard used in conjunction with NOTCE as follows: T ah is limited by the actual time required for handshaking and connection setup by the wireless protocol, while T ad is limited by the amount of information to be transmitted and by the actual data rate supported by the wireless standard. We also note that this conersion is necessary since ehicles trael at different speeds which are randomly distributed around some mean alue μ, and as a consequence the same distances r and 2r 2 will be coered by distinct ehicles in different time as follows: t ah is the actual time aailable for handshaking for a ehicle traeling at speed and represents the time in which it coers the radio range r of transceier#, while t ad is the actual time aailable for data exchange and for a ehicle traeling at speed represents the time in which the ehicle coers the radio range r 2 of transceier #2 (see Figures and 2). Their alues are gien by respectiely t ah = r = k T ah t ad = 2r 2 = k 2T ad n order for a ehicle in traffic to exchange information with the wireless sensor associated with a gien belt both stages of the communication process between ehicle and belt handshaking and data exchange must be completed successfully. We note that since these two stages are performed by two distinct radios mounted in opposite parts of the ehicle it is reasonable to assume that they are independent eents with probabilities P sh for successful handshaking, respectiely P sd for successful data exchange. Thus, the probability of successful information exchange P s between a ehicle and a sensor belt is gien by (3) (4) P s = P sh P sd (5) Successful handshking is defined by the condition t h t ah (6) where t h denotes the actual time required by a belt and a ehicle passing oer it to complete the handshaking stage and t ah is the time aailable handshaking gien in equation (3). We note that no actual data is exchanged by the belt and ehicle during handshaking and the time t h is essentially the hardware setup time required for setting up the physical radio link between belt and ehicle. Thus, the probability of successful handshaking is expressed as { P sh = Prob{t h t ah } = Prob k } T ah (7) t h and is gien by the cumulatie distribution function (CDF) F ( ) of the ehicle speed. Considering that is a Gaussian distributed random ariable [9], [] with aerage speed alue μ and ariance 2 equation (7) becomes ( kt ) ah t P sh = Q h (8) where Q( ) is the standard normal complementary CDF [, p. 4]. Once the handshaking stage has completed, the data exchange stage is successful if the amount of information to be exchanged between ehicle and belt is transferred while the ehicle is within the range 2r 2 of transceier #2. We note that is not the same for all ehicles since distinct ehicles may hae different information to exchange with the belt. Neertheless, as noted in [5], this amount is upper bounded, and in order to consider worst case scenario in our analysis we assume that is the maximum amount of information that a gien ehicle may exchange with the belt. The actual amount of data d that can be transferred between between ehicle and belt while they are within the communication range 2r 2 is gien by d = D t ad (9) where D is the data rate corresponding to transceier #2 expressed in [bits/s] and t ad is the time aailable for data exchange gien in equation (4). Thus, the probability of successful data exchange is expressed as { P sd = Prob{d } = Prob Dk } 2T ad () and is also gien by the cumulatie distribution function (CDF) F ( ) of the ehicle speed. With haing a Gaussian distribution with mean alue μ and ariance 2 equation () becomes ( ) D k2 T ad P sd = Q () Summarizing our results we hae that the probability of successful information exchange P s between a ehicle and a sensor belt is gien by [ ( kt )] [ ( )] ah t P s = Q h D k2 T ad Q (2) This alue implies that, out of the total number of ehicles that pass oer a gien belt, only a fraction η = P s will successfully exchange traffic-related information with the belt that can be aggregated and used for deciding whether a traffic incident has occurred. V. NUMERCAL RESULTS AND DSCUSSONS n order to ealuate the physical layer requirements for the NOTCE system we note that a alue of P s implies that the fraction of ehicles that pass oer a sensor belt and successfully exchange traffic-related information with it is η and implies an incident detection time of about minute or less as discussed in [8]. This alue for 5

4 P sh Aerage speed 45 mph Aerage speed 65 mph Aerage speed 75 mph P sh Aailable handshaking time 36 ms Aailable handshaking time 45 ms T ah [ms] Fig. 4. Probability of successful handshaking P sh as a function of the time aailable for handshaking T ah for fixed connection setup time t h =36ms and aerage ehicle speed equal to 45 mph, 65 mph, and 75 mph. P s can be achieed when both P sh and P sd, and in this section we inestigate scenarios which imply that these alues are achieed. We take the two conersion factors k = k 2 =65mph which is a common speed limit of most highways in the United States. For data exchange stage we assume that the maximum amount of information to be exchanged between ehicle and belt is 6 kbytes which is about 8 times more than the alue suggested in [5]. We first plot P sh ersus T ah for fixed t h = 36 ms and aerage ehicle speed equal to 45 mph, 65 mph, and 75 mph in Fig. 4 from where we note that for a gien alue of the aailable handshaking time T ah the probability of successful handshaking P sh decreases as the aerage speed of the ehicles increases. n order to achiee P sh = we need T ah 43 ms for aerage speed alue μ = 65 mph, T ah 32 ms for μ = 45 mph, and T ah 47 ms for μ =75mph. We also note that according to equation () an aailable handshaking time of 43 ms implies that the effectie handshaking range is about.25 m. Next we plot P sh ersus aerage ehicle speed μ for fixed aailable handshaking time T ah =36and 45 ms in Fig. 5 from where we note that, as expected, for a gien alue of the aerage ehicle speed the probability of successful handshaking decreases as the aailable handshaking time T ah decreases. We also note that in order to achiee P sh = the aerage ehicle speed is μ 52 mph for T ah =36ms and μ 68 mph for T ah =45ms. For the data exchange stage, we plot P sd ersus aerage ehicle speed μ for fixed aailable data exchange time T ad = 26 ms and data rate equal to D = 2, 3, and 4 kbps, in Fig. 6 from where we note that for a gien alue of the aerage ehicle speed μ the probability of successful data exchange decreases as the data rate decreases (an obious fact). n order to achiee P sd = with T ad = 26 ms the μ [mph] Fig. 5. Probability of successful handshaking P sh ersus aerage ehicle speed µ for fixed connection setup time t h = 36 ms and aailable handshaking time T ah =36and45ms. P sd.2. Data rate 2kbps Data rate 3kbps Data rate 4kbps μ [mph] Fig. 6. Probability of successful data exchange P sd ersus aerage ehicle speed µ for fixed aailable data exchange time T ad = 26 ms and data rate equal to D = 2, 3, and 4 kbps. aerage ehicle speed alue must be below μ 45 mph for data rate D = 2 kbps, respectiely μ 75 mph for data rate D = 3 kbps, while for data rate D = 4 kbps the data exchange is successful for any reasonable aerage speed of the ehicle. We also note that according to equation (2) an aailable data exchange time of 26 ms implies that r 2 =3m and the effectie range for data exchange is about 6 m which is consistent with the initial estimates in [5]. We also plot P sd ersus T ad for fixed aerage ehicle speed equal to 65 mph, data rate equal to D = 2, 3, and Speeds in excess of 75 mph are illegal on most highways in the United States. 6

5 P sd Data rate 2kbps Data rate 3kbps Data rate 4kbps probability of successful connection and information exchange around with radio transceiers that require approximately 4 ms (or less) for connection and operate at 3 kbps (or higher) data rates. These results indicate that inexpensie short-range wireless systems such as ZigBee [2] [4] are good candidates for the physical layer of the NOTCE system. n future work we plan to study in more detail the relationship between the parameters of arious wireless standards and the requirements of the NOTCE system identified in this paper in order to establish clearly which standards are suited best for this type of application T [ms] ad Fig. 7. Probability of successful data exchange P sd ersus aailable data exchange time T ad for fixed aerage ehicle speed equal to 65 mph, data rate equal to D = 2, 3, and 4 kbps. 4 kbps in Fig. 7 from where we note that for data rate equal to 3 kbps and aerage ehicle speed μ =65mph the probability of successful data exchange P sd = for T ad 93 ms. We also note that for aerage ehicle speed μ =65mph and data rate of 4 kbps the data exchange is successful for a wide range of alues for the aailable data exchange time alues while for the same aerage speed but data rate of 2 kbps the data exchange stage is essentially unsuccessful (with a probability around or below). V. CONCLUSONS We present an analysis of physical layer requirements for the NOTCE system proposed for notification of traffic incidents in ehicular communication networks [5]. The NOTCE system uses short-range wireless communication between ehicles and sensor belts embedded in the roadway in order to collect traffic information and to proide drier notification in case of traffic incidents. The limited transmission ranges of the radios inoled in the wireless communication process in NO- TCE impose constraints on time aailable for connection and data exchange, and we study the probabilities of establishing the wireless link and of successfully exchanging information between a sensor belt and a ehicle passing oer it. We derie expressions for the probability of successful connection setup (handshaking) and for successful information exchange as functions of seeral parameters such as the time aailable for handshaking/information exchange, aerage speed of the ehicle, data rate and amount of information to be exchanged between ehicle and belt, and we ealuate their alues for specific parameters corresponding to practical scenarios. Our numerical results show that for aerage ehicle speed of 65 mph (which is the legal speed limit on most highways in the US) a gien ehicle and sensor belt can achiee a ACKNOWLEDGMENT This work was supported in part by the National Science Foundation under Grant CNS REFERENCES [] ncident Management: Detection, Verification, and Traffic Management, US Department of Transportation, ntelligent Transportation Systems Field Operational Test Cross-Cutting Study, 998. [2] O. Ria and C. Borcea, The Urbanet Reolution: Sensor Power to the People, EEE Perasie Computing, ol. 6, no. 2, pp. 4 49, April-June 27. [3] O. Ria, T. Nadeem, C. Borcea, and L. ftode, Context-Aware Migratory Serices in Ad-Hoc Networks, EEE Transactions on Mobile Computing, ol. 6, no. 2, pp , December 27. [4] J. Nzounta, N. Rajgure, G. Wang, and C. Borcea, VANET Routing on City Roads Using Real-Time Vehicular Traffic nformation, EEE Transactions on Vehicular Technology, accepted for publication, to appear. Preprint aailable in EEE Xplore. [5] M. Abuelela, S. Olariu, and M. C. Weigle, NOTCE: An Architecture for Notification of Traffic ncidents, in Proceedings 65 th EEE Vehicular Technology Conference VTC 8 Spring, Singapore, May 28, pp [6] M. Karpinski, A. Senart, and V. Cahill, Sensor networks for smart roads, in PERCOMW 6: Proceedings of the 4th Annual EEE nternational Conference on Perasie Computing and Communications Workshops. Washington, DC, USA: EEE Computer Society, 26, p. 36. [7]. Chisalita and N. Shahmehri, On the Design of Safety Communication Systems for Vehicles, EEE Transactions on Systems, Man, and Cybernetics Part A: Systems and Humans, ol. 37, no. 6, pp , Noember 27. [8] D. B. Rawat, D. Treeumnuk, D. C. Popescu, M. Abuelela, and S. Olariu, Challenges and Perspecties in the mplementation of NOTCE Architecture for Vehicular Communications, in Proceedings 5 th EEE nternational Conference on Mobile Ad Hoc and Sensor Systems MASS 28, Atlanta, GA, September 28, pp [9] R. P. Roess, E. S. Prassas, and W. R. McShane, Traffic Engineering. Erewhon, NC: Pearson Prentice Hall, Third edition, 24. [] N. Wisitpongphan, F. Bai, P. Mudalige, V. Sadekar, and O. Tonguz, Routing in sparse ehicular ad hoc wireless networks, EEE Journal on Selected Areas in Communications, ol. 25, no. 7, pp , October 27. [] R. D. Yates and D. J. Goodman, Probability and Stochastic Processes. A Friendly ntroduction for Electrical and Computer Engineers, sted. New York, NY: John Wiley and Sons, 999. [2] H.-M. Tsai, O. K. Tonguz, C. Saraydar, T. Talty, M. Ames, and A. Macdonald, ZigBee-Based ntra-car Wireless Sensor Networks: A Case Study, EEE Wireless Communications, ol. 4, no. 6, pp , December 27. [3] A. Wheeler, Commercial Applications of Wireless Sensor Networks Using ZigBee, EEE Communications Magazine, ol. 45, no. 4, pp. 7 77, April 27. [4] R. Krastea, A. Bonea, G. Vesselin, and. Stoiano, Application of Wireless Protocols Bluetooth and ZigBee in Telemetry System Deelopment, Problems of Engineering, Cybernetics, and Robotics, ol. 55, pp. 3 38, 25, published by the nstitute of nformation Technology, Bulgarian Academy of Sciences, SSN