Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Download
Standard view
Full view
of .
Look up keyword
Like this
1Activity
0 of .
Results for:
No results containing your search query
P. 1
Mobility Prediction Progressive Routing (MP2R), a Cross-Layer Design for Inter-Vehicle Communication

Mobility Prediction Progressive Routing (MP2R), a Cross-Layer Design for Inter-Vehicle Communication

Ratings:

5.0

(2)
|Views: 39 |Likes:
Published by suhuatang4222
In this paper we analyze the characteristics of vehicle mobility and propose a novel Mobility Prediction Progressive Routing (MP2R) protocol for Inter-Vehicle Communication (IVC) that is based on cross-layer design. MP2R utilizes the additional gain provided by the directional antennas to improve link quality and connectivity; interference is reduced by the directional transmission. Each node learns its own position and speed and that of other nodes, and performs position prediction. (i) With the predicted progress and link quality, the forwarding decision of a packet is locally made, just before the packet is actually transmitted. In addition the load at the forwarder is considered in order to avoid congestion. (ii) The predicted geographic direction is used to control the beam of the directional antenna. The proposed MP2R protocol is especially suitable for forwarding burst traffic in highly mobile environments. Simulation results show that MP2R effectively reduces Packet Error Ratio (PER) compared with both topology-based routing (AODV, FSR) and normal progressive routing (NADV) in the IVC scenarios.
In this paper we analyze the characteristics of vehicle mobility and propose a novel Mobility Prediction Progressive Routing (MP2R) protocol for Inter-Vehicle Communication (IVC) that is based on cross-layer design. MP2R utilizes the additional gain provided by the directional antennas to improve link quality and connectivity; interference is reduced by the directional transmission. Each node learns its own position and speed and that of other nodes, and performs position prediction. (i) With the predicted progress and link quality, the forwarding decision of a packet is locally made, just before the packet is actually transmitted. In addition the load at the forwarder is considered in order to avoid congestion. (ii) The predicted geographic direction is used to control the beam of the directional antenna. The proposed MP2R protocol is especially suitable for forwarding burst traffic in highly mobile environments. Simulation results show that MP2R effectively reduces Packet Error Ratio (PER) compared with both topology-based routing (AODV, FSR) and normal progressive routing (NADV) in the IVC scenarios.

More info:

Published by: suhuatang4222 on Jan 06, 2009
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF or read online from Scribd
See more
See less

05/09/2014

 
IEICE TRANS. COMMUN., VOL.Exx–??, NO.xx XXXX 200x
1
PAPER
Mobility Prediction Progressive Routing (MP2R), aCross-Layer Design for Inter-Vehicle Communication
Suhua TANG
†∗
, Naoto KADOWAKI
††
,
and 
Sadao OBANA
,
Members
SUMMARY
In this paper we analyze the characteristics of vehicle mobility and propose a novel Mobility Prediction Progres-sive Routing (MP2R) protocol for Inter-Vehicle Communication(IVC) that is based on cross-layer design. MP2R utilizes theadditional gain provided by the directional antennas to improvelink quality and connectivity; interference is reduced by the di-rectional transmission. Each node learns its own position andspeed and that of other nodes, and performs position prediction.(i) With the predicted progress and link quality, the forwardingdecision of a packet is locally made, just before the packet isactually transmitted. In addition the load at the forwarder isconsidered in order to avoid congestion. (ii) The predicted geo-graphic direction is used to control the beam of the directionalantenna. The proposed MP2R protocol is especially suitable forforwarding burst traffic in highly mobile environments. Simula-tion results show that MP2R effectively reduces Packet Error Ra-tio (PER) compared with both topology-based routing (AODV[1], FSR [2]) and normal progressive routing (NADV [18]) in theIVC scenarios.
key words:
mobility prediction, position-based routing, cross-layer design, directional antenna, inter-vehicle communication 
1. Introduction
Mobile Ad hoc Networks (MANET) consist of mobilenodes cooperating to support multi-hop communica-tions. They are easy to deploy and may be used in manyfields such as Inter-Vehicle Communication (IVC). IVCbrings about new applications. One is extension of theInternet to the road, letting a vehicle access the Inter-net through the road-side Access Points (AP). This ischaracterized by the burst traffic. Another applicationmay be the continuous communication such as Voiceover IP (VoIP) between two vehicles moving along thesame trace. This may occur in the travel group of carsor the marching troop of tanks. When the distance be-tween a vehicle and an AP, or the distance between twovehicles, is beyond direct access, multi-hop communi-cation is necessary. In this paper the communicationwith up to 10 wireless hops is considered.Most of multi-hop routing protocols in MANETsare based on the global topology, where route calcu-
Manuscript received January 1, 2007.Manuscript revised January 1, 2007.Final manuscript received January 1, 2007.
ATR Adaptive Communications Research Laborato-ries, 2-2-2 Hikaridai “Keihanna Science CityKyoto 619-0288 Japan
††
Strategic Planning Department, NICT, 4-2-1 Nukui-Kitamachi, Koganei, Tokyo 184-8795 Japan
E-mail: shtang@atr.jp
lation depends on either proactive topology distribu-tion or on-demand route discovery. Typical proactiverouting protocols include distance vector protocols (e.g.DSDV) and link state protocols (e.g. FSR [2], OLSR).AODV [1] and DSR are known as on-demand routingprotocols, where a route is discovered when necessary.Some recent protocols also considered link quality in theroute calculation, either preferring strong links in theroute discovery stage [3]-[5], or making the initial routeconverge to the local optimum by gradually optimizingthe active route [6]. When the topology varies too fastwithout converging, performance of the topology-basedprotocols may be greatly degraded.On the other hand, in geographical routing proto-cols [8]-[13], with the position of the destination learnedin advance [14]-[15], each node can make the forwardingdecision locally. The greedy forwarding mode usuallyselects the longest link towards the destination. As aresult, the chosen link may have poor quality, and causefrequent retransmissions and packet loss. This is espe-cially obvious when gray zones [7] come into being intimes of fading. Then tradeoff should be made betweenlink quality and progress in selecting the forwardingnode [16]-[18].Some researchers also introduced directional an-tennas [20]-[21] into MANET to improve Signal to NoiseRatio (SNR) at the receiver and increase system capac-ity by space-division multiple access. Then one problemis to effectively determine the antenna beam ahead of actual transmission.In this paper we analyze the characteristics of ve-hicle mobility and show that neither conventional rout-ing protocols nor directional antenna work well in theIVC scenarios. Then we propose a Mobility-PredictionProgressive Routing (MP2R) protocol. With the cross-layer design, we try to jointly optimize routing, MAC,and beam control of directional antenna based on mo-bility prediction. In MP2R, the forwarder of a packetis selected according to link quality, load and progress.The local relative mobility is also taken into accountto opportunistically salvage a packet. And the actualtransmission is directional. Compared with the exist-ing MANET routing protocols, MP2R is more suitablefor the highly mobile IVC scenario.In the rest of the paper section 2 reviews the re-lated work. In section 3 we analyze the special charac-teristics of the IVC scenario and point out the research
 
2
IEICE TRANS. COMMUN., VOL.Exx–??, NO.xx XXXX 200x
problem. Then in section 4 we present the system ar-chitecture and the details of MP2R protocol. Section 5shows the evaluation results and the performance anal-ysis. Finally section 6 concludes the paper.
2. Related Work
2.1 Link Quality-aware RoutingToh reported the ABR protocol in [3], where each nodelearns the associativity of its neighbors by the periodicbeacons and utilizes it as the metric in route calcula-tion. In this way a route is discovered, preferring thenodes with less relative mobility. Dube et al. proposedin [4] the SSA protocol, which utilizes the signal sta-bility as the metric in the route discovery and preferslinks with high signal strength. Both protocols dependon the whole topology and can not well adapt to fastmobility in the IVC scenarios.De Couto et al. suggested the Expected Transmis-sion Count (ETX) metric [5], optimizing the routes byreflecting the real cost of each link in the route calcu-lation. However, it usually takes much time for ETXto converge, which limits its application in the mobilescenarios. In [6], the link metric is inversely associ-ated with Received Signal Strength Indication (RSSI).A route is maintained by the local route update scheme.It tracks link quality and topology variations in thepresence of low/moderate mobility.2.2 Position-based RoutingKo et al. proposed the LAR protocol [8] for the effi-cient route discovery, where the route request is onlyflooded in the specific zone estimated by the positionand moving speed of the destination. A typical geo-graphical routing protocol (GFG [9], GPSR [10]) uti-lizes the position information to forward data packetsinstead. When a packet is ready, a node chooses ina greedy way from its neighbors as the forwarder theone that is closest to the destination. This greedy for-warding mode may fail in the presence of local voidareas. Then the packet enters the perimeter mode andis forwarded by the right-hand rule in the planar graph[10]. Blazevic et al. proposed a more efficient schemefor routing around the void area in a large network,where packets are forwarded along a rough geographi-cal path specified by the anchors [13]. When a packetgets within two hops of the destination, its forwardingfalls back on the local link-state in order to compensatefor the position inaccuracy.The main demerit of position-based routing is ne-glect of link quality. The greedy forwarding mode tendsto result in transmission over long links with poor qual-ity. To solve this problem, Lee et al. proposed theNormalized ADVance (NADV) [18] metric, a combi-nation of the link quality aware routing and the geo-graphical routing. The link quality is reflected in thelink cost, which focused on Packet Error Ratio (PER).The greedy forwarding rule is embodied in the progress.Then the tradeoff is done by maximizing the progressnormalized by 1-PER. NADV is mainly proposed fordense networks where the nodes move at moderatespeeds and the greedy forwarding seldom fails. Sim-ilar to most of the local routing schemes, congestionmay occur at a node when many packets go through it.2.3 Utilization of Directional AntennaUtilization of a switch-beam antenna usually consistsof two steps. (i) Determining the beam forming di-rection to each neighbor. Bandyopadhyay et al. sug-gested building the Angle SNR Table (AST) for theElectronically Steerable Parasitic Array Radiator (ES-PAR) antenna [21]. In AST scheme each node peri-odically scans its neighbors as follows: An initiatorbroadcasts a SETUP frame by the omni-beam if thecarrier is sensed idle; this frame reserves the channel.Then the initiator broadcasts REQUEST frames witheach sector beam respectively, carrying the correspond-ing beam number. Finally the initiator transmits anEND frame by the omni-beam, releasing the channel.Then the neighbors of the initiator send back unicastREPLY frames indicating the heard beam numbers andthe associated RSSI, based on which the initiator learnsthe best transmission beam. (ii) Transmitting a dataframe with a suitable sector beam. If a valid cache of the direction to the peer node is found from the ASTtable, the DATA (or RTS) frame is transmitted withthe cached sector beam; otherwise the omni-beam isused instead. The ACK (or CTS) frame is transmittedwith the omni-beam. The antenna is immediately re-turned to omni-beam mode for the following receptionor carrier sense once the transmission is finished. Thereception is always under the omni-beam mode.Several problems exist in the AST scheme: (1) Theoverhead makes it not scalable to node density sinceeach node periodically performs beam scanning, whichinvolves a sequence of frame transmission at a relativelylow rate (one of the basic rates). (2) The instantaneousunstable RSSI due to fading degrades the beam stabil-ity. (3) The topology variation under high mobilityinvalidates the cached beam-forming direction quickly.2.4 Our MP2R SchemeCompared with most of the link quality aware rout-ing protocols (ABR [3], SSA [4], ETX [5], LHAOR[6]), MP2R takes the special characteristics of IVCinto account, where the forwarder selection is based onthe local topology and adapts better to relative mo-bility. MP2R reflects link quality in the forwarder se-lection, distinct from the position-based routing pro-tocols (LAR [8], GFG [9], GPSR [10]). Mobility pre-
 
TANG et al.: MOBILITY PREDICTION PROGRESSIVE ROUTING
3
diction is adopted in MP2R to mitigate the positioninaccuracy, in contrast with the mixed routing schemein TRR/TLR [13]. The progressive routing policy inMP2R is partially similar to NADV [18] with severalsignificant distinctions: (1) MP2R utilizes directionalantenna and calculates antenna beam by the positioninformation. (2) MP2R benefits from the relative mo-bility. When an intermediate node finds no forwardersfor a packet, it holds the packet and opportunisticallyforwards it later if a passing node comes near soon andreestablishes the connectivity. (3) The forwarding deci-sion of a packet in MP2R is made just before the packetis to be actually transmitted so as to reflect the poten-tial topology variation during its stay in the queue. (4)The load at the forwarding node is also considered inMP2R to distribute potential heavy traffic over multi-ple forwarders.
3. Characteristics of an IVC Scenario
In this section we analyze the characteristics of the IVCscenarios and define the research problem. An IVC sce-nario is quite different from a general ad hoc network.It mainly consists of three components: the roads, theintersections connecting roads, and the Mobile Nodes(MN). A road is usually bidirectional. MNs may runon either side of a road and MNs on the opposite sidesmove in the reverse directions. An MN may change itslane when overtaking others. Generally an MN onlychanges its moving direction at an intersection, whereit may be temporarily stopped by the signal.3.1 Characteristics of Vehicle CommunicationSpecifically, an IVC scenario has its special character-istics as follows.
Propagation model
. Since the obstructions sel-dom exist on the road and the Line-of-Sight (LOS) pathusually exists, the radio signal generally propagates interms of the two-ray model, which depends on the dis-tance between two MNs. The LOS propagation meansthat the transmission direction to a neighbor matchesthe geographic direction.
Regular speed
. Each MN in its normal statealmost runs at a constant speed on a lane with a rel-atively stable direction. Therefore the position of anMN in the near future is predictable.
Node density
. To guarantee the safety, vehi-cles moving on the same lane should be separated farenough so that an urgent brake will not make themcollide with each other. Then among the MNs alongthe same direction sometimes only few nodes are withinthe communication range. To improving the connectiv-ity MNs in the opposite direction can be used. Mean-while the directional antennal can also be applied. Itincreases the connectivity and reduces interferences bythe directional transmission.
SBDP
1
ACEFP
2
P
3
P
4
SBDP
1
ACEFP
2
P
3
P
4
Fig.1
Problem Definition.
Topology and route variation
. Several reasonsresult in quick topology variations. Firstly, among theMNs in the same direction an MN may overtake theothers and the high relative speed between two MNs inthe reverse directions results in short-term links whichbreak soon. The transmission direction changes greatlywhen two MNs cross each other, which makes it difficultto control the antenna beam. Secondly, the forwardingpath tends to vary greatly near an intersection. Whena group of MNs gets near to the intersection they maydecrease the speeds. Their distance gets short and theroute shortens. The route stretches again as the groupof MNs leaves the intersection. Sometimes a suddenchange of the signal at an intersection may even split agroup of MNs into two disconnected parts.
Potential variation of routes
. A packet maywait in the queue due to congestion or in the presenceof burst traffic. In the highly mobile environment, asthe packet is actually transmitted the best route mayhave already changed and differ from the one that wasdecided when the packet entered the queue.3.2 Problem DefinitionFigure 1 shows a typical IVC scenario where each MNis equipped with a directional antenna. MNs
,
,
D
moves in the same direction and
is communicatingwith
D
. If there is an MN at
1
,
can use it to forwardpackets to
, which further forwards to
D
. When thisMN does not exist, the route may temporarily breaks.Instead
may opportunistically use the passing MN
B
as the instantaneous forwarder. Then two problemsoccur: (1) How does
detect the link to
B
quickly atthe high relative speed? (2) How does
B
control itsantenna beam to
? When
B
moves from
3
to
4
and
moves from
1
to
2
, the transmission directionfrom
B
to
changes over 90 degree.The special characteristics of the IVC scenario re-quire a new routing protocol to forward packets underthe high mobility and an efficient beam control schemefor the directional antenna. With the cross-layer designand mobility prediction, all MNs, including those witha high relative speed in the opposite direction, are usedto forward packets. The forwarding decision made atthe time a packet is to be actually transmitted furtheravoids the potential route variation when the packetstays in the queue. And the position-based beam con-trol scheme effectively controls the directional antenna.

You're Reading a Free Preview

Download
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->