In-car shopping: A data dissemination scheme for vehicular networks

0 downloads 0 Views 356KB Size Report
we proposed a new data dissemination scheme in this paper. Our scheme ... use a large-scale communication infrastructure as an .... The format of Interest is.
In-Car Shopping: A Data Dissemination Scheme for Vehicular Networks in Urban Areas Chen Chen, Cong Wang, Guoxian Zhang, Zhiyuan Ren

Jingjing Ma Railway Campus, Central South University, Changsha, 410075, China

State Key Laboratory of Integrated Services Networks, Xidian University, Xi’an, 710071, China [email protected] Abstract—Vehicular Ad hoc Network(VANET), an infusive area for the development of Intelligent Transportation System (ITS), is going to be a significant communication pattern in our moving life. Data dissemination is an important part of VANETs. The demand of disseminate commercial advertisements in moving cars emerges. To satisfy the demand of shopping in car, we proposed a new data dissemination scheme in this paper. Our scheme aims at disseminating the shopping information cached in RSUs. We adopt the Information Centric Network (ICN) in forward mechanism and take both pull-based and push-based in to account. At the same time, our communication model combines V2V and V2I. In this way, our scheme improves the performance of traditional data dissemination and makes in-car shopping a reality. In addition, in order to avoid broadcast storm and the waste of resources, Distance Based Forwarding (DBF) is employed in broadcasting when vehicle desire a content unit. A computing method of XOR is also adopted in RSUs to avoid information redundancy. Simulation results show that the proposed protocol outperforms existing protocols in terms of the request success ratio, request-response delay, message overhead and collision rate. Keywords—Vehicular ad-hoc network; ICN; data disseminate; DBF algorithm; forward mechanism

I. INTRODUCTION Data dissemination services embrace a wide variety of telematics applications where data packets are generated from a remote server and destined to a group of nomadic users such as vehicle passengers and pedestrians. In VANETs, typical applications of data dissemination services include traffic information downloading, entertainment content distribution, and commercial advertising. In this article, to meet the everincreasing demands for in-car shopping, a data dissemination scheme is proposed. As far as we know, due to some inherent characteristics of VANETs, such as the dynamic of topology, frequent network disconnections and varying network densities, the data communication is a challenging and indispensable task actually. Data dissemination is one of the challenging research topics in this area. In general, there are two basic strategies for data dissemination in VANETs. The first one is the push-based approach. In this approach, the source node disseminates the message proactively broadcast to other nodes. According to

this approach, many vehicles can receive the message, however, some vehicles may do not need the message, so this approach may waste the communication resources. The other is the pull-based approach, in which vehicles send requests to obtain the messages that they want, and the vehicles that have the relevant messages reply to the requests. This approach satisfies the requester, but it is not easy because of the weak connection between vehicles, what’s more, in our scene of incar shopping, the vehicles need to obtain the push messages in terms of the interest of some merchants. So we use a new kind of infrastructure to meet the ever-increasing demands of drivers and merchants through absorbing the merits of both pull and push-based approaches in this paper. The VANETs provides two categories of data communication which named vehicle-to-infrastructure communication (V2I) and vehicle-to-vehicle communication (V2V). In this paper, we combined the two communication ways to increase the request success ratio of the vehicles. We use a large-scale communication infrastructure as an information resource to receive requests and disseminate data because of its high capacity and data processing ability. The infrastructure can be named Road Side Unit (RSU) which is similar with the Road Side Buffer (RSB)[1]. The RSBs are lightweight and low-cost devices which have the limited buffer storage and is able to transmit the cached contents to fast-moving vehicles wirelessly. The RSU we proposed can also deal with vehicles’ requests by a computing method of XOR to solve the problem of information redundancy. In our paper, not only RSUs can receive requests and disseminate data to vehicles, but also other vehicles can forward requests and reply. We broadcast the requests by distance based forwarding (DBF). The basic principle of DBF is that the forwarding vehicle is picked as the one farthest away from the sender among all those vehicles who have receives a new message waits for a defer timer which decreases with the sender-receiver distance before resend the message[2], this algorithm can solve the problem of broadcast storm and avoid incurring high overhead. As described above, we apply a communication model named “information-centric” to improve the performance of data dissemination [3]. Different from “host-centric” model which establishes host-based and end-to-end communications, “information-centric” model focuses on the delivery of named data, and data are cached at

978-1-5386-2062-5/17/$31.00 ©2017 IEEE

routers along the forwarding path (such as the forwarding vehicles in this article). Consequently, in terms of delay, our scheme outperforms the tradition data disseminate protocol. Our contributions are summarized as: x

We propose a new data dissemination scheme to realize “in-car shopping” effectively. The strategy can achieve the goal that disseminate commercial information with higher request success ratio and lower delay.

x

We take both pull and push-based dissemination into account, which can reduce the redundant information while ensuring each node gets the desired information.

x

We combine two kinds of data communication (V2V and V2I) in “information-centric” communication model to adapt the characters of VANETs while considering realistic data dissemination.

The rest of this paper is organized as follows: the related works is given in Section II. In section III, we state our system model and assumptions. In section IV, the new protocol is proposed and analyzed, section V gives the simulation results. Finally, section VI provides a conclusion. II. RELATED WORKS Data dissemination in VANET has attracted lots of interests in the past few years, many strategies for data dissemination in vehicular environment have been suggested. Most of these strategies can be classified in push and pull approaches which have been mentioned in section I. However, according to the strategies’ characteristics, these strategies can be further classified as follows. Some clustering methods suited for vehicular networks have been proposed. Nitin MASLEKAR, etc. proposed that clusters are formed by driving direction in the intersection in [4], and the strategy they proposed only disseminates information to the vehicles which are moving along the same direction. Simulations show that it relieves the traffic load in the network, but it cannot be ensured that the vehicles’ needs can be satisfied. In [5], authors proposed an information dissemination scheme based on the autonomous clustering and P2P network to disseminate the information message in the VANET environment. The vehicles’ connectivity is intermittent because of the fast moving of vehicles. Researchers have proposed the disruption tolerant networks (DTN), which relays on the opportunistic dissemination technique. In this communication system, some technique follows a store and forward method: the nodes not only receiving and forwarding the information, also store the information in a buffer, carry it along until they come in contact with other nodes. This model is usually referred to the story-carry-forward model, e.g. in [6], vehicles stored data and forwarded it by different schemes. Authors in [7] introduced an epidemic protocol for information dissemination in the context of DTN, it calculated the dissemination probability by collecting requests from different directions, but this scheme didn’t take the request success ratio into consideration.

Recently, some researchers have focus on content-based dissemination [8-10], which is based on information-centric model, the data are cached at routers or forward nodes along the forwarding path, which can be used to satisfy future requests. Researchers take the geographical position, encounter probability or some else factors into account to decide how to disseminate data. However, those paper only consider the inter-vehicles communication, in our paper, we take V2I into account at the same time. Actually, the main ideal of our paper is inspired by Mohammad Ali Maddah-Ali[11]. He proposed that the server could send XOR signal to tell users to exchange data they need if each user has one subset of the file that the other user needs. In VANET, the nodes can move fast, so we suppose that the RSUs can compete for the job of XOR, the data exchange between vehicles can be an auxiliary method. III. SYSTEM MODEL/ASSUMPTIONS In this section, the system model and assumptions are presented as follows: A. System model We consider the scene as a bounded road of a city which is shown in Fig.1. The infrastructures around the grocery and supermarket separately are named RSU which is equipped with a wireless transceiver operating on the dedicated shortrange communication radio and can communicate with nearby vehicles using the vehicle-to-infrastructure. A case in point, the RSU can disseminate the advertisements of some goods in the supermarket periodically, and the vehicles in the RSU’s communication range can not only receive these information, but also send requests to the RSU to obtain the data they want, such as some discount information. What’s more, the vehicles out of the communication range of RSU can require and obtain data by the help of relay vehicles. The details of our forwarding protocol will be described in section IV.

Supermarket

Grocery

Fig. 1. The scene of system model

For the sake of redundancy reduction, we assume that every content unit has a Time-to-Live value (TTL), the nodes will drop data when their TTL expires. Then, we introduce some definitions and notations that will be used throughout the remaining of the paper. Definition 1. Service: The message that a consumer desires is called service in this paper. It is composed of different business. We describe the service A as

A{A1 , A2 ,... Ai ,...(i 1, 2,...)}. Definition 2. Business: The business includes the video data, audio data or some pictures, they consist of content units which are composed of the most basic information unit of information — bit, a case in point, a business Ai is composed of a sequence of content units Ai1 , Ai 2 ,..., Aij ,... i, j

1, 2,... .

and the content unit packet denoted as (seq, data), the seq represent the sequence number of the content unit, such as A11,A12,…, the data represents the content unit. We describe the business Ai as A^ Ai1 , Ai 2 ,..., Aij ,... i, j

1, 2,... ` ,

Definition 3. Interest: When a vehicle want content unit, the node will send an Interest. The format of Interest is denoted by (seq, R-ID, all-ID). The “seq” is the serial number of the content unit, R-ID is the ID of the consumer and all-ID is the ID of all the nodes which have send or forwarded the interest. Definition 4. Caching list (CL): The caching list records the serial number of the content units stored in the node. Definition 5. Pending interest list (PIL): The pending interest list stores all the Interest that a node has forwarded but not satisfied yet. B. Assumptions To make our work practical, the assumptions are made as follows: i) The vehicles drive at a constant speed, the radio range and cache capacity are identical, they have the ability to compose content units into service. ii) Every vehicle sends out hello message with a mean frequency of 1 Hz at random instants to obtain neighbors’ information. iii) Nodes in the VANET can play three different roles: x x

x

They send an interest when they desire content unit. If there is no RSU in the interested node’s communication range They will meet other nodes’ interests if they have the content units the interested node needs They can help the requesters acquire content units if they could not satisfy the request nodes. IV. PROPOSED PROTOCOL

Our protocol is introduced in this section, it includes the broadcasting mechanism and how to send the interest/content unit. The flow chart of the whole protocol is shown in Fig.2. A consumer has to broadcast an interest when it wants to obtain a service. However, in order to avoid the problems of broadcast storm and the waste of resources, we adopt a forwarding algorithm of DBF which describe in part A. If a node wants some information, the node will send an interest, we describe the interest forward mechanism in part B. Then if a node receives the interest and has the matching content unit in its buffer and there is no RSU in the communication range of forwarder, the node would send the content units back, otherwise, the RSU will deal with interests by XOR and send

the matching content units back. The details of forwarding mechanism of content units and the method of XOR are shown in part C and D, respectively. 7KHFRQVXPHUZDQW WRREWDLQVHUYLFH

3UHSDUHWRVHQWDQLQWHUHVW )LJ

7KHFRQVXPHUJHWWKHFRQWHQW

3UHSDUHWRVHQGWKHPDWFKLQJ FRQWHQWXQLW )LJ

Fig. 2. Flow chart of the protocol

A. Using DBF in broadcasting We use the DBF algorithm to avoid the broadcast storm when senders forward the interests and make use of the channel resource effectively. The main idea of DBF algorithm is relaying packets only by the receiver who is farthest from the sender in its communication range. In order to find the farthest node, vehicles should set a defer timer when it receives the interest. The longer the distance between sender and receiver, the shorter the defer time. When defer time expired, the node will forward the interest. If a node receives the same interest during the time, that means the interest has been forwarded by the farther node, the vehicle will discard the information and abrogate the timer. The most important is how to set the value of each vehicle’s timer. The timer is decided by the distance between sender and receiver, and it is inversely proportional to the distance, in the other words, the vehicle which is farther from the information sender have a shorter defer time. The expire time is computed by (1).

Rd ]uT , (1) S where R is the wireless transmission radius. d is the distance from the previous forwarder. S is the step length, which controls the effect of the defer time. And T=R/c, where c is the speed of light. Tslots

[

B. The forwarding of the interest If a vehicle wants to obtain some services, it will broadcast its interests. When the vehicle can communicate with the RSU directly, it sends the interest to the RSU. Otherwise, the vehicle need to acquire response with other nodes’ help. When the consumer receives the matching content unit, it will broadcast an acknowledge message to inform its neighborhoods to clear their defer timer, otherwise, the neighbor node who has the smallest value of defer timer will forward interest after the timer expires. When a vehicle is about to send or forward an interest, the vehicle needs to perform the following steps. The flow chart of preparing to send an interest is described in Fig.3.

1) The vehicle looks up its own caching list to find whether there is the matching content unit in its own cache buffer to satisfy the interest. If cache hit, the node will get ready for sending the matching content unit. Otherwise 2) If there is a RSU in the vehicle’s communication range, the vehicle will send the interest to the RSU directly. Otherwise 3) The node locating its PIL to find whether the seq in the interest packet has been recorded. If the desired content has been recorded and the R-ID in interest packet is not same with the seq which are recorded, the vehicle adds the R-ID of new interest packet into the PIL. Then the sender take method of DBF forwards this interest. Otherwise 4) The vehicle adds an entry in the PIL.

recorded in all-ID. Then delete the matching entry in PIL. 3UHSDUHWRVHQGWKHFRQWHQW XQLW

7KHVHTLV UHFRUGHGLQ 3,/"

1R

'LVFDUG