|عنوان فارسی مقاله:||اولین تجربه با CarTorrent در بستر آزمون شبکه ادهاک خودروی حقیقی|
|عنوان انگلیسی مقاله:||First Experience with CarTorrent in a Real Vehicular Ad Hoc Network Testbed|
|رشته های مرتبط:||مهندسی برق، مهندسی فناوری اطلاعات، فناوری اطلاعات و ارتباطات، اینترنت و شبکه های گسترده، مخابرات سیار، سامانه های شبکه و شبکه های مخابراتی|
|فرمت مقالات رایگان||مقالات انگلیسی و ترجمه های فارسی رایگان با فرمت PDF میباشند|
|کیفیت ترجمه||کیفیت ترجمه این مقاله پایین میباشد|
مقاله انگلیسی رایگان
|دانلود رایگان مقاله انگلیسی|
ترجمه فارسی رایگان
|دانلود رایگان ترجمه مقاله|
|جستجوی ترجمه مقالات||جستجوی ترجمه مقالات|
بخشی از ترجمه فارسی مقاله:
بخشی از مقاله انگلیسی:
Content sharing using cooperative peer-to-peer model has become increasingly more popular in a vehicular ad hoc network (VANET). The small transmission window from a vehicle to an access point (AP), high mobility of vehicles, and intermittent and short-lived connectivity to an AP provide incentives for vehicles to cooperate with one another to obtain information from the Internet. These characteristics of VANETs naturally stipulate the use of cooperative peer-to-peer paradigm and motivate related content sharing application such as CarTorrent. Building upon previous research on SPAWN[6, 1], we have implemented CarTorrent and deployed it on a real VANET. We have run extensive field tests to affirm the feasibility of the peer-to-peer file sharing application tailored to VANET. To the best of our knowledge, the deployment of such a content sharing application on a real vehicular ad hoc testbed is the first of its kind.
Navigation safety requirements have propelled the development and deployment of VANET. Beyond the safety navigation, new types of applications have emerged such as office-on-wheels and on-car entertainment. Among these, file sharing is gaining its momentum: People want to download not only music and movie trailers while driving, but also location-cognizant data such as virtual hotel tour clips. People can download files from road-side access points (APs) that provide Internet connections, which is known as Wardriving . The conventional client-server model will not work neither scale well for the following reasons. First, due to the high mobility, the actual contact time to an AP is short. For example, assuming that the WiFi range is 300m, when driving at the speed of 45mph, we can have 30 seconds of contact period. With the overhead of association, DHCP, and Internet connections, the actual contact period is shorter than 30 seconds. Second, in real environments signal strength is mainly a function of distance; i.e., as the distance from the AP increases, the signal strength decreases. This increases the packet error rate; consequently, the effective throughput that one can achieve is much less than expected. Third, it is neither practical to install APs every 300 meters, nor feasible to stop in the middle of roads to download a file. Thus, we conclude that in reality, the contact period is short, and its goodput is low. To effectively handle this situation, we advocate the use of peer-to-peer file swarming in which users out of AP range can still download parts of files from others. In P2P file swarming such as BitTorrent, a file is divided into the same size pieces, and peers with fractions of a file can exchange whatever pieces available by forming an overlay network. This not only sheds the load of the server, but also increases the availability of pieces, thus expediting the downloading process. However, BitTorrent cannot be directly ported to wireless environments because of the discrepancy between a logical overlay topology and a physical topology of mobile nodes. For instance, a peer who is one hop away in a logical overlay could be located five hops away physically. To maximize the available wireless resources by localizing traffic (i.e., which increases the spatial diversity and reduces the routing overheads), researchers thus far have focused on mapping the logical overlay to physical topology [3, 7, 1]. In particular, SPAWN uses the proximity-driven piece selection strategy, thus further reducing the average hop count of multi-hop pulling. It is known that the proximity-driven piece selection outperforms the conventional “rarest first” piece selection. In this paper, we propose CarTorrent, a BitTorrent-style file swarming protocol in the vehicular environment, by extending SPAWN . For a given file, CarTorrent clients disseminate their piece availability information via gossip- ing (i.e., by k-hop limited scope broadcasting). Each gossip message is forwarded until it reaches to nodes located k-hop away from the originator. Thus, peers can gather statistics such as local topology and piece availability. Statistics are then used to select a piece/peer that is preferably close in proximity. In other words, given that two peers A and B own a rarest piece that C desires, C would choose A because A has a shorter hop count to C than B does. CarTorrent users can send queries to other clients; the query is delay tolerant such that whenever the connectivity is available the query is sent out and resolved. Note that the craving for downloading information, e.g., sightseeing landmarks, movie previews, outweighs the penchant for keeping to oneself and thereby provides incentives for cooperative content sharing. The goal of this paper is to test the feasibility of in-vehicle content sharing (i.e., CarTorrent). Toward this goal we implement CarTorrent and measure its performance in a real VANET testbed. The use of the peer-to-peer application on a real VANET testbed is the first of its kind. We show that peers can utilize the gossip mechanism to recognize one another’s presence and employ the piece-selection strategy to optimally download files from one another. We run extensive field tests and obtain performance measurements in a real VANET testbed. We demonstrate performance comparisons between baseline static parking lot and real road mobile scenarios. We believe that many lessons learned and technologies picked up to set up a testbed of this size are invaluable to VANET research. The rest of this paper is organized as follows. Section 2 illustrates CarTorrent’s architecture and implementation details. Section 3 presents our experiment setup and results. Section 4 shows the related work. Finally, Section 5 draws conclusion of the paper, and present the possible future work.