Hindawi Publishing Corporation International Journal of Distributed Sensor Networks Volume 2014, Article ID 650923, 8 pages http://dx.doi.org/10.1155/2014/650923
Research Article Social Group Architecture Based Distributed Ride-Sharing Service in VANET Weicheng Zhao, Yajuan Qin, Dong Yang, Linjuan Zhang, and Wanting Zhu Beijing Jiaotong University, Beijing 100044, China Correspondence should be addressed to Weicheng Zhao;
[email protected] Received 4 November 2013; Accepted 15 January 2014; Published 16 March 2014 Academic Editor: Deyun Gao Copyright © 2014 Weicheng Zhao et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Lots of traditional distributed ride-sharing services match vehicles with drivers and passengers geographic information only. When urban road congestion situation is particularly serious, such as Beijing’s rush hour, the waiting time for the passengers is too long which affects the quality of the service. So we propose a distributed ride-sharing service based on dual Social Group Architecture (SGA). We divide the service into Drivers Social Group Architecture (DSGA) message and Vehicles Social Group Architecture (VSGA) message. Vehicles generate dual SGA messages to complete ride-sharing service. DSGA messages focus on the relation between drivers and passengers. We make a basic geometry matching by generating the DSGA messages. VSGA messages figure out the traffic conditions through a multilevel detection. After generating VSGA messages, the final matching strategy is processed. The low level detection is finished by the limited neighbors to decrease the network consume. Dual SGA messages shorten the waiting time for the passengers and avoid traffic jams. Analysis shows that our scheme enhances the ride-sharing service quality and robust.
1. Introduction Due to the increasing growth in the number of vehicles in urban roads, the vehicles have played a great role in our lives. And vehicular ad hoc network (VANET) has been proposed. Vehicle installed with some on-board sensors to exchange neighbor nodes information is a principal part of the VANET communications. The infrastructure like Road Side Unit (RSU) is also an optional facility in the VANET. Two mainly communication ways are listed as follows: vehicles to the infrastructures (V2I) and vehicles to the vehicles (V2V) mentioned as intervehicular communications. Considering the RSU is not possible to fully deploy in the urban roads; the intervehicular communications which help to complete all kinds of information service experience are a hot spot in VANET. Nowadays, much attention has been focused to the research in the field of vehicular information service. Information service tremendously facilitates our quality of traveling life by sharing the entertainment information and improving the driving efficiency. Ride-sharing service is most
widely accepted and used by public and the government. The car sharing service is proposed to reduce the use of private cars which obviously reduces the traffic jams and saves fuel energy. This service brings a lot of economic and environmental influence to our urban roads. Lots of information service methods have been created to deal with it. Ride-sharing service is the update of the traditional street hail service [1]. In China lots of software algorithm based on the cell phone access to internet has been used in urban vehicles. Such taxi sharing service depends on the mobile phone and needs installing the same third-part software. Such kind of service has two obvious weaknesses. (1) The data copyright is dominated by the internet network operator which may restrict the free development of the ride-sharing service. (2) This service is a request-driven only method. Vehicle drivers need to pay much attention to the software and respond as soon as possible, which may distract drivers and make them in danger.
2
International Journal of Distributed Sensor Networks
So the car sharing service based on the dedicated VANET way such as Dedicated Short Range Communications (DSRC) is more attractive [2]. Lots of the services based on a centralized infrastructure, which needs a lot of labor and resources to manage and maintain, are hard to expand. According to that, we study and analyse a distributed ridesharing service algorithm. Unlike a centralized car sharing system, realtime is the most significant character of this distributed network system. Vehicles and pedestrians can exchange messages directly, and thus they bring some most striking characteristics like decentralization, low-cost management, and robustness. Traditional distributed services concentrate on raising the utilization ratio of seat to save fuel and reduce traffic jam. All the methods mention the relationship between drivers and passengers to calculate whether this ride-sharing service is convenient and fuel saving in a geometry way. But when the traffic condition from the vehicles location to the passengers location is terribly jammed, such methods will encounter three troubles. (1) After replying the request, the vehicles may get idle along the more congested path to pick passengers. Thus, they will waste a lot of time and cost additional fuel. (2) When pedestrians receive the request reply, they have to wait for a long time and they have no idea about the detailed waiting time. (3) Because of the long time waiting and high cost, both the vehicles and the passengers may get angry and even give up this ride-sharing deal. So the service quality and the matching reliability are very poor. To solve these problems, in this paper, we propose an efficient car-sharing service based on Dual Social Group Architecture (SGA). We divide the service into Drivers Social Group Architecture (DSGA) identifier and Vehicles Social Group Architecture (VSGA) identifier. The former concerns about the drivers and the pedestrians relations while the later focuses on the vehicles driving area. We first collect our DSGA messages which contain the current location of the request publisher and the drivers, the distance between the drivers and passengers destination, and so on. With that we make a basic matching to determine whether to request in a similar direction. Then, we will do a best forwarding strategy and transfer the message to the relay vehicles. Meanwhile, all relay vehicles will generate their own VSGA message to reflect surrounding traffic conditions. And if the basic matching is successful, the vehicles will converge all of them and generate the VSGA information. Such information can figure out the detailed road conditions from the current vehicles location to the passengers. And with a threshold comparator, only if the waiting time is acceptable, the final matching completes. So the basic matching ensuring the ride-sharing is fuel saving in geometry way, and final matching avoids the traffic jam in our ride-sharing service. The own VSGA message generation is a crucial part in our DSGA framework; we should improve the matching precision and, meanwhile, do not add to much network bandwidth and operation cost. So we propose a multilevel
jam detection way to detect road congestion which contains both feature level fusion and decision level fusion. The relay vehicles first give some tokens to their neighbour nodes. And the neighbour who has the token will do a feature level detection, collecting the atomic road condition messages (contains the speed, accelerator, and brake frequency) within the scope of communication and with a fuzzy clustering-based message aggregation approach to decrease the information redundancy by extracting abnormal information. Then the relay vehicles will receive all the token neighbours message. Finally, the relay vehicles finish the decision level detect with a basic probability assignment based information fusion (BPAIF) method and generate their own VSGA messages. To reduce network overhead, we embed all of the atomic messages into the beacon packet. The rest of the paper is organized as follows. After talking about related work in Section 2, we give the main framework of our dual-SGA based distributed information service in Section 3. Section 4 introduces how the DSGA message generates and the basic ride-sharing request matching strategy. Section 5 presents the VSGA identifier based information generation and final matching make. The benefit analysis will be well discussed in Section 6. We conclude the contribution in Section 7 finally.
2. Related Work Domestic structures and international institutions have done some research about the carpooling service [3–5]. Many scholars begin to study mobile phone-based carpooling service software. Drivers need to keep an eye on the software and respond to the request as soon as possible. This method affects traffic safety and adds extra network traffic cost. Reference [6] proposed a EZCab car-sharing system in dedicated VANET. It gives four kinds of information transfer algorithms in vehicles routing. But the algorithm is too complex and network overhead is too heavy. Reference [7] simplified the algorithm, but it requires the vehicles to restore lots of historical information. Reference [8] uses RSU to complete the taxi sharing service, while it is not impossible to be fully deployed in urban road. Reference [9] proposed a centralized ride-sharing system which provides two types of riding-sharing services. One will require the users to make register in prior and the other one makes a regular route for the car sharing service. The ride-matching process needs to be operated at least once a day. Thus non-real-time attribute is not suitable for the fastpaced carpooling service nowadays. Reference [10] proposed a distributed ride-sharing system. It deploys the RSU in the urban road. The RSU periodically collects and broadcasts the carpooling information within its hop communication range. Following that, vehicles evaluate and plan a carpooling service travel path in a geometric method. However, if lots of vehicles stay in the same carpooling RSU center, the fuel-saving ability and matching success rating performance will decrease rapidly. Reference [11] introduces a carpooling way to save fuel and reduce pollution. By calculating the current and destination location, finding a most convenient path, the authors plan the travel path in advance. All of
International Journal of Distributed Sensor Networks the methods focus on the passengers number and people’s destination, marked as DSGA message in our scheme. They do not consider the traffic conditions between the passengers and the vehicles. So if the vehicles get stuck in a jam road, the waiting time and the fuel consumption will increase enormously. And the success matching rate will decrease rapidly because of the traffic jam. In this paper we proposed an efficient distributed carpooling service system based on Dual-SGA. It generates the DSGA message in a geometric method and detects the traffic conditions stored in the VSGA message. The own VSGA message generation to detect traffic jam is a crucial part in our DSGA framework. Most popular traffic jam detection can be divided into two formats, the low-level feature information detection and the high-level decision level information detection. The low-level information aggregation method employs feature messages to tremendous atomic messages. Threshold-based message aggregation and road segment-based message aggregation are the most classical and well-known schemes. Traffic view system [12] is a good paradigm, which contains two fusion rules, minimum aggregation cost based and distance based to build message pairs. While, the scheme performs a low efficiency in reality, the number of message pairs satisfying the rules is limited. Wischof proposed a method to enhance the fusion productivity named as SOTIS traffic information system [13]. The fusion standard is the road segmentation here. However, the information precision and exactitude are influenced heavily by the length of the road segmentation. In [14], the paper puts to use a cluster-based message fusion methodology to reduce the information redundancy named as StreetSmart system. While considering the algorithm demands completely connected network topology, it is not compatible with the highly dynamic and intermittently connected characters of the VANET. To reduce maintenance overhead, [15] put forward a fuzzy logic-based structurefree approach. As our preliminary work, [16] proposed information fusion way, but it cannot be applied to the VANET as no detailed supported background scene. A structurefree information fusion approach is proposed in [17]. It uses a multiple level parallel fusion method to strength the reliability of message aggregation, but it is only suitable for the information fusion of vehicular nodes located within the fixed distance of an exist event such as vehicle collision. But in the realistic scenario, the road congestion area of influence is highly dynamic with various times. In [18], Ramon Bauze puts forward the CoTEC (Cooperative Traffic congestion detection) approach to sense the traffic jam with a fuzzy logic-based information fusion methodology. However, the constrained bandwidth resources will be rapidly exhausted by large number of redundant message.
3. System Model First of all, a well-designed Vehicle Ad Hoc Network (VANET) model is defined for this travel information service. VANET consists by large numbers of vehicles equipped with high quality sensor nodes.
3 In this paper, the authors consider a new and highly efficient ride-sharing service based on Social Group Architecture (SGA). SGA is a novel conception and angle of view to classify and solve information service problems in VANET. According to the Complexity Theory, a social group network will be established if we regard each member in the group as a node, the relationship between different members as an edge. We survey the inner relations and analyse the Social Group Architecture of vehicles and drivers in VANET. If we make vehicles as nodes, regard the driving areas as edges, the vehicles based social group architecture is formed. We mark it as VSGA identifier. While regarding drivers as nodes, the points of areas as edges and drivers based social group architecture (DSGA) are formed, mentioned as another SGA identifier. In ride-sharing service, the messages are divided into the DSGA based message and VSGA based message. We mark the vehicles’ traffic conditions as the VSGA information and regard the destinations of drives’ geographic position as VSGA message. The DSGA message enhances the utilization of the vehicles while various pedestrians can take same vehicles to a similar destination. And VSGA message helps the vehicles find a better path to reduce the time cost of waiting in the road. Figure 1 shows our basic information service system model; the neighbour vehicles mean the neighbour of the token vehicles. Pedestrians who desire traveling to some places publish their current geographical location information and destination geographical location information as a ride-sharing request. With the device supporting the 802.11p protocol, the generated messages are delivered to the relay transportation. The latter one received the request and detected related traffic surroundings. Following that, with DSGA identifier, each vehicle processes the DSGA message generation procedure and makes a basic matching in a geometric method. If the vehicles are not matched, we do the optimization request message forwarding according to the value of utility. With VSGA identifier, each vehicle sensor detects and collects surrounding information; through a multilevel message detection process, the traffic conditions message is formed which describes the road surroundings specifically and efficiently. Finally the DSGA matched relay vehicles weight all DSGA information from the vehicles to the passengers, and the vehicle will respond to the pedestrians if the messages are ultimately matched. Otherwise, they will transmit the carpooling messages to other relay vehicular nodes until the request is ultimately satisfied.
4. DSGA Message Generation and Basic Matching Method 4.1. DSGA Message Generation. DSGA identifier message analyzes the role which drivers and passengers play in this problem. In this paper, we consider the ride-sharing information service beginning when passengers publish a ride-sharing request and ending in the pedestrians’ acknowledgement. As shown in Figure 2 The ride-sharing request information transmits by selected relay vehicles until the
4
International Journal of Distributed Sensor Networks Forwarding
Relay vehicles
Neighbor vehicles Atomic VSGA messages
Optimal forwarding No
No
Basic matching
Yes
Beacon interaction
Final matching
DSGA message
VSGA information
Calculate Geographic information
Decision level message
Generate
Request publish
Token release
Token vehicles Feature level message
Reply request
Figure 1: System framework.
Passengers
Token release Request Devices support 802.11p
Drivers
Token vehicles Request
Relay vehicles Reply
Reply
Sensors equip in vehicles
Figure 2: Information matching steps.
drivers match the message. Following that, the drivers reply the message with the same route strategy while the message contains the driver’s geographical location and arriving time. Finally passengers received the first reply and answer with acknowledgement. We make some basic settings to better describe our ridesharing service framework. No infrastructures are employed in the Roadside. So the service information transmits only from vehicle to vehicle (V2V). Both the vehicles and the pedestrians are equipped with the devices which support the 802.11p protocols and Global Position System (GPS). But the information service is stored in a limited cache buffer, so the service messages are deleted periodically according to the lifetime. We assume that both public transport means and private cars are playing a crucial role in the information service deliver. To facilitate the vehicles transmit the information actively, we provide some privileges for the vehicles which contribute a lot to the information service network. So the motivated vehicles which are engaging in the message forwarding possess the superior bandwidth resource and more message forwarding priority. Taxi and private vehicles are both involved in this ride-sharing service. In the DSGA identifier message, each vehicle contains two message lists the vehicles information list and publish message list. The drivers information DI includes {𝐷ID , 𝐷type , 𝐷𝐿 𝑡−1 , 𝐷𝐿 𝑡 , 𝐷𝐷, 𝐷𝑡 }, which represent, respectively, as follows driver identifier, vehicle type, driver geographical location at time 𝑡 − 1, driver geographical location at time 𝑡, passenger destination (equals 0 when empty state), and message time to live. The publish information PI reflects sequentially as follows: publisher
identifier, request vehicles type, passenger geographical location, passenger destination location, request time to live, and request generation time. In addition, the location contains the longitude and latitude coordinates (𝑥, 𝑦), where 𝑥 means longitude and 𝑦 means latitude. 4.2. Basic Geometry Matching Method. When the relay vehicles received the request, the DSGA message will be first calculated to math with the request. If the math processes succeed, the vehicles will give a reply to the pedestrians. Otherwise, the message will be transmit to next relay vehicles. The math processes is as follows. (a) Determine whether the vehicle driving is close to the passengers: (𝑑𝐷𝐿−1 𝑃𝐿 − 𝑑𝐷𝐿 𝑃𝐿 ) > 0.
(1)
If the equation is satisfied, which means carpooling service is possible, go to step (b). Otherwise, it will quite the process. (b) Determine inequality 𝑑𝐷𝐷 𝑃𝐷 ≤ 𝑑̂𝐷𝐷 𝑃𝐷 , where 𝐷𝐷 denotes the destination of the driver and 𝑃𝐷 refers to the destination of the passengers. 𝑑𝐷𝐷 𝑃𝐷 reflects the direct distance of two destination positions. [𝑑̂𝐷𝐷 𝑃𝐷 ] figures out the threshold of the destination similarity. If parameters satisfy the inequality, the direction to the destination is similar. And we finish our match. Otherwise, we continue our matching procedures. (c) Finally we estimate the related direction of the destinations. According to Figure 3, 𝑃𝐿 𝑃𝐷 means the direction vector from current location 𝑃𝐿 to the destination 𝑃𝐷 of the passengers. In the same way, 𝐷𝐿 𝐷𝐷 refers to the direction vector from current location to the destination of the driver. 𝜃 denotes the intersection angel between 𝑃𝐿 𝑃𝐷 and 𝐷𝐿 𝐷𝐷. We can derive that cos 𝜃 =
2 𝑑𝑃2𝐿 𝐷𝐷 + 𝑑𝑆2𝐿 𝑆𝐷 − 𝑑𝐷 𝐷 𝑃𝐷
2𝑑𝑃𝐿 𝐷𝐷 ⋅ 𝑑𝑃𝐿 𝑃𝐷
,
(2)
International Journal of Distributed Sensor Networks
5
DD (x2 , y2 )
DD (x2 , y2 )
DL (x1 , y1 )
DL−1 (x0 , y0 ) DL (x1 , y1 )
𝜃 𝜃 1 d⊥ PL (x3 , y3 )
a
Token vehicles Relay vehicles Passenger
where 𝑑𝑃𝐿 𝐷𝐷 , 𝑑𝑃𝐿 𝑃𝐷 , and 𝑑𝐷𝐿 𝐷𝐷 separately describe the distance among the current location of the passengers or driver and the destinations of the passengers and driver. The value of cos 𝜃 ranges from 0 to 1; the higher value means the related direction of passengers and vehicles is more similar. Apparently we can derive another equation, 𝑑⊥ = 𝑑𝐷𝐿 𝑃𝐿 ⋅ √1 − cos2 𝜃1 = 𝑑𝑃𝐿 𝑃𝐿
2 2 2 𝑑𝐷 + 𝑑𝐷 − 𝑑𝐷 𝐿 𝐷𝐷 𝐿 𝑃𝐿 𝐷 𝑃𝐿
2𝑑𝐷𝐿 𝐷𝐷 ⋅ 𝑑𝐷𝐿 𝑃𝐿
PL (x3 , y3 )
PD (x4 , y4 )
Figure 3: Basic geometry matching.
⋅ √1 − (
A
2
(3)
),
where 𝑑⊥ denotes the perpendicular distance between the passengers location and vector 𝑉𝐿 𝑉𝐷. We predefine two thresholds 𝜃̂ and 𝑑̂⊥ to classify the angle 𝜃 and vertical distance 𝑑⊥ . If 𝜃 < 𝜃̂ and 𝑑⊥ < 𝑑̂⊥ , the vehicles are convenient to take the passengers and VSGA messages math succeed.
5. VSGA Information Generation and Final Matching Algorithm 5.1. VSGA Framework. VSGA information reflects the whole path traffic condition from passengers to the matching vehicle. The routing procedure is shown as Figure 4. Each relay vehicle collects and fuses their own VSGA information with a multilevel massage gathered way. For one relay vehicle we give 𝑛 tokens to control the number of the decisionlevel meta data. The relay vehicles transmit the tokens in the beacon packet to their neighbors. And the neighbor vehicles with the tokens will first collect the basic feature-level atomic messages with the one-hop scope of the communications. And then the vehicles will fuse them with a fuzzy cluster way to generate the feature-level result. Following that, the vehicle returns the tokens and delivers the feature-level result to the relay vehicles. Finally the relay vehicles make the decision-level information fusion and generate its own VSGA identifier information. The matching vehicles received all the VSGA identifier information from relay vehicles which reflect the price and detailed path traffic conditions.
PD (x4 , y4 )
Token neighbor Other vehicles
Figure 4: VSGA information matching structure.
The detailed multilevel message aggregation procedure presents the architecture of the feature level aggregation FCMA and the decision level aggregation BPAIF. The main purpose of the low level fusion is to decrease the huge information redundancy and abstract the crucial traffic jam evidence for high level fusion. The high level one is responsible for making a decision of the traffic jam. According to the FCMA, vehicles with in-board sensors will first accumulate their neighbor atomic messages they have met and then restore them in the message cache. After a period of time, the fuzzy clustering module uses the direct fuzzy clustering method depending on their similar relations. Atoms of the information in the same cluster will converge and generate feature information. It will be disseminated to the neighbor vehicle nodes when the gathered feature message is unusual. To get a high accuracy and efficiency in traffic congestion detecting, fake and useless feature messages need to be eliminated to avoid a misjudgment which may aggravate the traffic congestion. In the BPAIF, the basic probability assignment (BPA) is derived by the credibility message assignment and event probability prediction. Later, the traffic congestion evidence given by multiple BPA values converges to one ultimate BPA value by the D-S evidence combination criterion. Then, depending on our modified BPAIF, the final decision-making vehicle will employ the ultimate BPA value to detect the traffic congestion event. If the incident of traffic jam being judged finally, vehicle nodes will disseminate the traffic congestion message by broadcast at once. 5.2. VSGA Message Detection in Token Vehicles. As shown if Figure 5, the own VSGA message generation is another key part in our DSGA framework; we should improve the matching precision and, meanwhile, do not add to much network bandwidth and operation cost. So, we proposed a multilevel jam detection way to detect road congestion which contains both feature level fusion and decision level fusion. The relay vehicles first give some tokens to their neighbour nodes. And the neighbour who has the token will do a feature
6
International Journal of Distributed Sensor Networks
A
Token E
Token vehicles
Token neighbor
Relay vehicles
Other vehicles
equals 5 minutes, in normal crossroads it is referred to as 10 minutes, and at main road crossroads we set it to 12 minutes. We add an extra time condition on the BPA computation in (4). Assuming that ∃𝐴, 𝐵 ⊂ 𝑈, if 𝑚(𝐴), 𝑚(𝐵), and 𝑚(𝑈) satisfy the following conditions, the event “A” will be the result of decision: max {𝑡(𝑛𝑖 ) } − min {𝑡(𝑛𝑖 ) } ≥ 𝑇 , 1≤𝑖≤𝑠 𝑟 wait 𝑟 1≤𝑖≤𝑠 𝑚 (𝐴) − 𝑚 (𝐵) > 𝜀1 ,
(4)
𝑚 (𝑈) < 𝜀2 ,
Figure 5: Traffic congestion happened on two-way road.
𝑚 (𝐴) > 𝑚 (𝑈) , level detection, collecting the atomic road condition messages (contains the speed, accelerator, and brake frequency) within the scope of communication and with a fuzzy clusteringbased message aggregation (FCMA) approach to decrease the information redundancy by extracting abnormal information. Then the relay vehicles will receive all the token neighbours message. Each relay vehicles will generate the VSGA information by using multilevel detections. Relay vehicles release token to the token vehicles. Token vehicles collect the atomic message within communication range and do a fuzzy clustering-based message fusion. All relay vehicles receive all the token vehicles fusion message to generate their own VSGA messages. The VSGA messages can figure the traffic conditions in double vehicles communication distance. If basic geometric matching succeeds, the VSGA message will converge into VSGA information. VSGA information guides the vehicles to complete the final request matching. In VANET, vehicular state massages are generated by onboard sensors, such as vehicle speed 𝑠, accelerator 𝑒, brake frequency 𝑏, horizontal and vertical geographical coordinate (𝜎, 𝜌), and message creation time 𝑡, which can reflect the road traffic conditions. These pieces of information can be gathered into one atomic message and we call them message attributes of the atomic message. For instance, the atomic message 𝑘 of vehicle node 𝑖 is denoted by 𝑥𝑘(𝑖) = (𝑠𝑘(𝑖) , 𝑒𝑘(𝑖) , 𝑏𝑘(𝑖) , (𝜎𝑘(𝑖) , 𝜌𝑘(𝑖) ), 𝑡𝑘(𝑖) ), where 𝑠𝑘(𝑖) , 𝑒𝑘(𝑖) , 𝑏𝑘(𝑖) , (𝜎𝑘(𝑖) , 𝜌𝑘(𝑖) ), and 𝑡𝑘(𝑖) denote the attributes of message 𝑘, respectively. 5.3. Final Decision Strategy. The relay vehicles will receive all the token neighbours message and finish the decision level detect with a basic probability assignment based information fusion (BPAIF) method and generate their own VSGA messages. To reduce network overhead, we embed all of the atomic messages into the beacon packet. If basic geometric matching succeeds, the VSGA message will converge into VSGA information. VSGA information guides the vehicles to complete the final request matching. The VSGA information reflects the whole path traffic conditions. So we can advancingly calculate the waiting time with a BPA evidence decision. We derive our final decision strategy as follows. Where 𝑇wait refers to the threshold of waiting time, here we use three different thresholds to describe the latency time 𝑇wait depending on the density. Where in a T-junction the 𝑇wait
where 𝜀1 , 𝜀2 are predefined threshold values.
6. Analysis We will analyse our Dual-SGA identification efficiency in the car sharing service. Our simulation is constructed by two parts. The first part is the VSGA-based message generation efficiency. VSGA-based message generation is a great novel contribution to enhance our carpooling service matching success rate. And it is also a very efficient information gathering way in spite of the ride sharing. So we will analyse it to demonstrate high accuracy. The second part discusses the dual-SGA based identification improving the matching ratio in carpool service. As traditional ride-sharing service only focuses on the drivers willing, obviously our scheme performs much better than the traditional way. In the feature level, the relay vehicles give some limited tokens to the token vehicles. And the token vehicles will collect feature-level rough message. By exchanging and signaling with the cover communication neighbors in beacon packet, the token vehicles complete the fuzzy clusteringbased information fusion. Then the relay vehicles will receive the feature level detection result. In the feature level, lots of the token neighbors send packet consuming the network bandwidth. But we embed it into beacon packet interaction which ensures the information accuracy while reducing network traffic load. We analyse the basic performance of our system. Passengers publish request and reply with acknowledgement to the VANET. The moving vehicles, as the real vehicles moving base on the roadmap, find the appropriate way to take the passengers, finding the destination and loop again. The static vehicles (passengers) will publish a request every 60 seconds. We set the speed in the free road as 60 KM/H. And we speed down to the 30 km/H in the slight road congestion segment. Also, we have a one discount as 3 KM/H speed in the road congestion path. And the distance between vehicle and passengers is 50 KM. Our SGA scheme will only go to the free road. And DSGA only scheme will go to the free road, slight road congestion segment, and road congestion segment with the probability 𝑃 = 1/𝑁, where 𝑁 denotes the number of the request. So we calculate the waiting time by 𝑡 = 𝑠/V, where 𝑠 refers to the distance 50 KM and V refers to the speed in the road.
International Journal of Distributed Sensor Networks
7 complete the final request matching. Analysis shows that the VSGA information is quite efficient and does not add much to network expenses. It enhances the ride-sharing service quality and robust. The SGA scheme also shortens the waiting time for the passengers and avoids traffic jams.
20 18 16 14 12 10 8 6 4 2 0
Conflict of Interests The authors declare that there is no conflict of interests regarding the publication of this paper. 10
20 30 40 50 The number of published requests
60
DSGA + VSGA DSGA
Figure 6: Waiting time for each passenger.
Figure 6 shows the waiting time for passengers meeting their couple vehicles after replying the ACK. The horizontal axis shows the different number of request in an hour and the accomplished request number shows in the vertical axis. The dark one shows the DSGA+VSGA scheme waiting time for the passengers meeting the matching vehicles, while the light way finds the DSGA scheme waiting time. We can easily find that the Dual-SGA is more than DSGA only in matching and greatly shortens the waiting time for the passengers. SO, our scheme will perform much better than the DSGA scheme. The waiting time will cost a lot of gas and time on the congestion road, while the passengers also can be agitated that “I got a matching car, but I do not know where it is.” Such situation will tremendously decrease our carpooling service quality and reliability. It obviously proves that our Dual-SGA scheme is much more precise and fits the passenger request. Especially for class in the morning and evening rush, the traffic jam ubiquitously exists.
7. Conclusions In this paper, we propose a distributed carpooling service method based on dual Social Group Architecture (SGA). By giving the access vehicles two SGA identifiers in VANET, vehicles will generate DSGA message which focus on the relation of drivers and passengers and VSGA messages which figure out the traffic conditions. DSGA messages contain the pedestrians and vehicles destination geographic information which gives a basic geometry matching. If match fails, the DSGA message is forwarded to the relay cars. Meanwhile each relay vehicle will generate the VSGA message by using multilevel detections. Relay vehicle releases token to the token vehicles. Token vehicles collect atomic messages within communication range and do a fuzzy clusteringmessage fusion. Each relay vehicle receives the token vehicles fusion message to generate its own VSGA messages. The VSGA messages can figure out the traffic conditions in double vehicles communication distance. If basic geometric matching succeeds, the VSGA message will converge into VSGA information. VSGA information guides the vehicles to
Acknowledgments The authors gratefully acknowledge the supports of the National Natural Science Foundation of China (NSFC) (no. 61272504), Specialized Research Fund for the Doctoral Program of Higher Education (no. 20130009110010), Beijing Natural Science Foundation (no. 4122057), National S&T Major Program (no. 2012ZX03005003), and the Fundamental Research Funds for the Central Universities (no. 2014JBM006).
References [1] M. L. Macy, S. J. Clark, G. L. Freed et al., “Carpooling and booster seats: a national survey of parents,” Pediatrics, vol. 129, no. 2, pp. 290–298, 2012. [2] N. Agatz, A. Erera, M. Savelsbergh, and X. Wang, “Optimization for dynamic ride-sharing: a review,” European Journal of Operational Research, vol. 223, no. 2, pp. 295–303, 2012. [3] D. Mukherjee, S. Banerjee, and P. Misra, “Ad-hoc ride sharing application using continuous sparql queries,” in Proceedings of the 21st International Conference Companion on World Wide Web (WWW ’12), pp. 579–580, ACM, Lyon, France, April 2012. [4] F. Drews and D. Luxen, “Multi-hop ride sharing,” in Proceedings of the 6th Annual Symposium on Combinatorial Search, 2013. [5] P. M. d’Orey, R. Fernandes, and M. Ferreira, “Empirical evaluation of a dynamic and distributed taxi-sharing system,” in Proceedings of the 15th International IEEE Conference on Intelligent Transportation Systems (ITSC ’12), pp. 140–146, IEEE, Anchorage, Alaska, USA, September 2012. [6] P. Zhou, T. Nadeem, P. Kang, C. Borcea, and L. Iftode, “EZCab: a cab booking application using short-range wireless communication,” in Proceedings of the 3rd IEEE International Conference on Pervasive Computing and Communications (PerCom ’05), pp. 27–40, IEEE, Kauai Island, Hawaii, USA, March 2005. [7] J.-P. Sheu, G.-Y. Chang, and C.-H. Chen, “A distributed taxi hailing protocol in vehicular ad-hoc networks,” in Proceedings of the IEEE 71st Vehicular Technology Conference (VTC ’10-Spring), IEEE, Taipei, Taiwan, May 2010. [8] N. Liu, M. Liu, J. Cao, G. Chen, and W. Lou, “When transportation meets communication: V2P over VANETs,” in Proceedings of the 30th IEEE International Conference on Distributed Computing Systems (ICDCS ’10), pp. 567–576, IEEE, Genoa, Italy, June 2010. [9] C. Tao and C. Wu, “Behavioral responses to dynamic ridesharing services—the case of taxi-sharing project in Taipei,” in Proceedings of the IEEE International Conference on Service Operations and Logistics, and Informatics (IEEE/SOLI ’08), vol. 2, pp. 1576–1581, IEEE, Beijing, China, October 2008.
8 [10] A. Arora, M. Yun, T. Kim, Y. Zhou, and H.-A. Choi, “Automated ride share selection using vehicular area networks,” in Proceedings of the IEEE International Conference on Communications Workshops (ICC ’09), pp. 1–6, IEEE, Dresden, Germany, June 2009. [11] P.-Y. Chen, J.-W. Liu, and W.-T. Chen, “A fuel-saving and pollution-reducing dynamic taxi-sharing protocol in VANETs,” in Proceedings of the IEEE 72nd Vehicular Technology Conference Fall (VTC ’10-Fall), IEEE, Ontario, Canada, September 2010. [12] S. Dietzel, F. Kargl, G. Heijenk, and F. Schaub, “Modeling in-network aggregation in VANETs,” IEEE Communications Magazine, vol. 49, no. 11, pp. 142–148, 2011. [13] S. Roy, R. Sen, S. Kulkarni, P. Kulkarni, B. Raman, and L. K. Singh, “Wireless across road: Rf based road traffic congestion detection,” in Proceedings of the 3rd International Conference on Communication Systems and Networks (COMSNETS ’11), pp. 1– 6, Bangalore, India, January 2011. [14] R. Sen, P. Siriah, and B. Raman, “Roadsoundsense: acoustic sensing based road congestion monitoring in developing regions,” in Proceedings of the 8th Annual IEEE Communications Society Conference on Sensor, Mesh and Ad Hoc Communications and Networks (SECON ’11), pp. 125–133, Salt Lake City, Utah, USA, June 2011. [15] D. N. Georgiou, T. E. Karakasidis, J. Nieto, and A. Torres, “Use of fuzzy clustering technique and matrices to classify amino acids and its impact to Chou’s pseudo amino acid composition,” Journal of Theoretical Biology, vol. 257, no. 1, pp. 17–26, 2009. [16] L. Zhang, D. Gao, W. Zhao, and H.-C. Chao, “A multilevel information fusion approach for road congestion detection in VANETs,” Mathematical and Computer Modelling, vol. 58, no. 5-6, pp. 1206–1221, 2013. [17] L. S. Buriol, M. J. Hirsch, P. M. Pardalos, T. Querido, M. G. Resende, and M. Ritt, “A biased random-key genetic algorithm for road congestion minimization,” Optimization Letters, vol. 4, no. 4, pp. 619–633, 2010. [18] F. Terroso-S´aenz, M. Vald´es-Vela, C. Sotomayor-Mart´ınez, R. Toledo-Moreo, and A. F. G´omez-Skarmeta, ““A cooperative approach to traffic congestion detection with complex event processing and VANET,” IEEE Transactions on Intelligent Transportation Systems, vol. 13, no. 2, pp. 914–929, 2012.
International Journal of Distributed Sensor Networks
International Journal of
Rotating Machinery
Engineering Journal of
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
The Scientific World Journal Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
International Journal of
Distributed Sensor Networks
Journal of
Sensors Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Journal of
Control Science and Engineering
Advances in
Civil Engineering Hindawi Publishing Corporation http://www.hindawi.com
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Volume 2014
Submit your manuscripts at http://www.hindawi.com Journal of
Journal of
Electrical and Computer Engineering
Robotics Hindawi Publishing Corporation http://www.hindawi.com
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Volume 2014
VLSI Design Advances in OptoElectronics
International Journal of
Navigation and Observation Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Hindawi Publishing Corporation http://www.hindawi.com
Chemical Engineering Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Volume 2014
Active and Passive Electronic Components
Antennas and Propagation Hindawi Publishing Corporation http://www.hindawi.com
Aerospace Engineering
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Volume 2014
International Journal of
International Journal of
International Journal of
Modelling & Simulation in Engineering
Volume 2014
Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Shock and Vibration Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
Advances in
Acoustics and Vibration Hindawi Publishing Corporation http://www.hindawi.com
Volume 2014
