Broadcasting Methods in Mobile Ad Hoc Networks An Overview

Broadcasting Methods in Mobile Ad HocNetworks:An Overview∗Demetres Kouvatsos∗and Is-Haka Mkwawa+∗Department of Computing,University of Bradford,UK.+Department of Computer Science,University College Dublin,Ireland.∗D.D.Kouvatsos,+I.M.Mkwawa@AbstractOptimal broadcasting in mobile ad hoc networks is crucial for pro-viding control and routing information for multicast and point to pointcommunication algorithms.This tutorial presents an overview on the state of the art of broadcast-ing techniques in mobile ad hoc networks and makes recommendations toimprove the efficiency and performance of the current and future broad-casting techniques.1IntroductionA mobile ad hoc network(MANET)is a special type of wireless mobile network which forms a temporary network without the aid of an established infrastruc-ture or a centralised administration.The applications of MANETs range fromthe civilian use to emergency rescue sites and in battlefield.Each node in MANET is a router.If a source node is unable to send a message directly to its destination node due to limited transmission range,the source node uses intermediate nodes to forward the message towards the destination node.The main challenges in MANET are reliability,bandwidth and battery power.The network has unpredictable characteristics,it’s topology,signal strengthsfluctuates with environment and time,communication routes breaks and new ones are formed dynamically.In this context,communication algorithms and protocols should have very light in computational and storage needs in order to conserve energy and bandwidth(c.f.,[1–5]).∗This work is partially supported by the EU NoE Euro-NGI.T09/1Broadcasting is the process in which a source node sends a message to all other nodes in MANET.Broadcasting is important in MANET for routing information discovery,for instance,protocols such as dynamic source routing (DSR)[6],ad hoc on demand distance vector(AODV)[7],zone routing pro-tocol(ZRP)[2,8,9]and location aided routing(LAR)[10]use broadcasting to establish routes.Broadcasting MANET poses more challenges than in wired networks due to node mobility and scarce system resources.Because of the mobility there is no single optimal scheme for all scenario.This tutorial is organised as follows,Section2reviews the probabilistic and de-terministic broadcasting methods.Section3and4review the cluster and tree based broadcasting methods,respectively.Tree based an cluster based broad-casting methods improvements are included in Section5and6,respectively. Conclusions follow in the last Section7.The reviewed broadcasting methods are clearly depicted in Figure1.2Broadcasting MethodsBroadcasting methods have been categorised into four families utilising the IEEE802.11MAC specifications[11].Note that for the comparisons of these categories the reader is referred to[12].1.Simpleflooding[13,14],requires each node in a MANET to rebroadcastall packets2.probability based[15],assigns probabilities to each node to rebroadcastdepending on the topology of the network3.area based[15],common transmission distance is assumed and a node willrebroadcast if there is sufficient coverage area and4.neighbourhood based[16–20],State on the neighbourhood is maintainedby neighbourhood method,the information obtained from the neighbour-ing nodes is used for rebroadcastApart from simpleflooding,each broadcasting category aims at optimising en-ergy and bandwidth by minimising message retransmission.Each category is described in the following subsection.2.1Simple Flooding MethodIn this method,a source node of a MANET disseminates a message to all its neighbours,each of these neighbours will check if they have seen this message before,if yes the message will be dropped,if no the message will re-disseminated at once to all their neighbours.The process goes on until all nodes have the message.Although this method is very reliable for a MANET with low densityT09/2broadcastdeterministicself−pruningscalable broadcastingprobalistic distance−basedcounter−basedlocation−basedad hoc broadcastingsimple floodingcluster−basedFigure 1:Broadcasting methodsnodes and high mobility but it is very harmful and unproductive as it causes severe network congestion and quickly exhaust the battery power.A polynomial number of messages is of the magnitude (n 2)in a MANET of size n and is depicted in Figure 2.2.2Probability Based Methods 2.2.1Probability Based ApproachThe probability based approach tries to solve the problems of the simple flood-ing method.Each node i ∈N is given a predetermined probability p i for re-broadcasting.In this context,having some nodes not to rebroadcast min-imises the network congestion and collisions.In this approach there is a danger that some nodes will not receive the broadcast message.∀i s.t.p i =1,theT09/30 204060 80100 0 2 46 8 10N u m b e r o f M e s s a g e s Number of Nodes messages Vs nodesx**2-xFigure 2: (n 2)number of messages in a simple flooding methodprobability based approach is reduced to a simple flooding approach.More effi-cient broadcasting reduces p i as the number of neighbour density increase and vise versa.2.2.2Counter-Based Scheme ApproachIn this approach the random assessment delay (RAD)is set,a threshold K is determined and a counter k >=1is formed on the number of times the broadcast message is received.During the RAD,the counter k is incremented by one for each redundant message received and if k >K when RAD expires,the message is dropped.Otherwise,it is rebroadcast.In this approach,some nodes will not rebroadcast in a more dense MANET while in a less dense MANET all nodes will rebroadcast.2.3Area Based MethodThis method is comprised of distance and location based approaches.2.3.1Distance Based ApproachIn Section 2.2.2the counter is used to decide either to drop a message or to rebroadcast,in this section a distance between a receiving node and it’s neigh-bours will decide between the two.Let d be the distance between the receiving node and the source node,if d is very small then the rebroadcast coverage of the receiving node is also very small.If d is large then the rebroadcast coverageT09/4is large.If d=0then the rebroadcast coverage is0too.A receiving node will normally determine the threshold distance D and set the RAD and redundant messages will be stored until the RAD expires.When RAD expires,all distances from source nodes will be checked,if d<D then the received messages will be dropped,otherwise the messages will be rebroadcast. Ni et al[15]has suggested that the signal strengths can be used to calculate the distance from the source node.The role of distance can even be directly replace by the signal strength by setting the signal strength threshold.2.3.2Location Based ApproachIn this approach each node must have the means to establish it’s own location in order to estimate the additional coverage more precisely.This approach can be supported by global positioning system(GPS)[21].Each node in a MANET will add its own location to the header of each message it sends or rebroadcasts.When a node received a message,thefirst thing will be to note the location of the sender and compute the additional coverage area to rebroadcast.If the additional coverage area to rebroadcast is less than the given threshold,the message is dropped when the RAD expires,otherwise the message will be rebroadcast.The problem of the location based approach is the cost of calculating additional coverage areas,which is calculating many intersections among many circles. This will drain the scarcely available energy.2.4Neighbour Knowledge Method2.4.1Self PruningEach node in this approach is required to have knowledge of it’s neighbours, this knowledge can be achieved by periodic”Hello”messages.The receiving node willfirst compare its neighbour lists to that of sender’s list,the receiving node will rebroadcast if the additional nodes could be reached,otherwise the receiving node will drop the message.This is the simplest approach in the neighbour knowledge method.In Figure3,after receiving a message from node 2node1will rebroadcast the message to node4and node3as it’s only additional nodes.Note that node5also will rebroadcast the same message to node4as it’s only additional node.In this situation still the message redundancy takes place.2.4.2Scalable Broadcasting ApproachThis approach improves the self pruning approach as the chances of message retransmission are still higher in the later approach.T09/5324567 1Figure3:Self pruning approachThis approach requires that all nodes in a MANET have knowledge of their neighbours up to a two hop distance.This knowledge is established by”Hello”messages.In this approach,each node has a two hop topology information.In Figure3,node1receives a message from node2,since node2is a neighbour, node1has knowledge of all its own and node2neighbours who have received the broadcast message.The additional nodes of node2will receive the message rebroadcast by node2with the aid of RAD.Note that still node4will receive the redundant message.Pen and Lu[18]dynamically adjusted RAD according to a given MANET conditions.Each node will look for a neighbour with maximum degree in it’s knowledge base.When a neighbour with the maximum numberdegreeδi is found,the RAD is computed based on the ratioδiδ,whereδis thedegree of the current node.Nodes with large RAD will always rebroadcastfirst.2.4.3Ad Hoc Broadcasting ApproachIn this approach,only nodes selected as gateway nodes and a broadcast message header are allowed to rebroadcast the message.The approach is described as follow,1.locate all two hop neighbours that can only be reached by a one hopneighbour.select these one hop neighbours as gateways.2.calculate the cover set that will receive the message from the currentgateway set3.for the neighbours not yet in the gateway set,find the one that wouldcover the most two hop neighbours not in the cover set.Set this one hop neighbour as a gateway.T09/64.repeat process2and3until all two hop neighbours are covered.5.when a node receives a message and is a gateway,this node determineswhich of its neighbours already received the message in the same transmis-sion.These neighbours are considered already covered and are dropped from the neighbour used to select the next hop getways.In Figure4,node2has1,5and6nodes as one hop neighbours,3and4nodes has two hop neighbours.Node3can be reached through node1as a one hop neighbour of node2.Node4can be reached through node1or node5as one hop neighbours of node2.Node3selects node1as a gateway to rebroadcast the message to nodes3and4.Upon receiving the message node5will not rebroadcast the message as it is not a gateway.324516Figure4:Ad Hoc broadcast approach2.5Shortcoming of Existing Broadcasting MethodsThe shortcomings are deduced from a detailed comparative studies in[12].1.all methods apart from neighbour based methods require more rebroad-casts,with respect to the number of retransmitting nodes[15].2.methods that make use of RAD suffer in high density MANETs unless amechanism to adapt a nodes RAD to its local environmental behaviour are developed.3.because it does not use local information to decide whether to rebroadcastor not,the ad hoc broadcasting approach have difficulties in a very high mobile MANET.Base on the comparative studies by[12],none of the existing broadcasting pro-tocols are satisfactory for wide ranging MANET environments.T09/7Because of its adaptive nature,scalable broadcast approach has significant im-provements over the non adaptive approaches.Due to these shortcomings there is a need to develop new efficient broadcasting approaches with the common goal of conserving the available scarce resource in MANETs.3Cluster Based MethodsThe methods highlighted in Section2are based on statistical and geometrical models which estimate the additional coverage of re-broadcasting,in this section clustering based methods,based on graphic theoretic concepts are introduced.Note that the clustering approach has been used to address traffic coordination schemes[23],routing problems[23]and fault tolerance issues[24].Note that cluster approach proposed in[15]was adopted in[22]in order to reduce the complexity of the storm broadcasting problem.Each node in a MANET periodically sends”Hello”messages to advertise its presence.Each node has a unique ID.A cluster is a set of nodes formed as follows.A node with a local minimal ID will elect itself as a cluster head.All surrounding nodes of a head are members of the cluster identified by the heads ID.Within a cluster,a member that can communicate with a node in another cluster is a gateway.To take mobility into account,when two heads meet,the one with a larger ID gives up its head role.This cluster formation is depicted in Figure5.HX XHX GatewayH HeadFigure5:Clustered MANETNi et al[15]assumed that the cluster formed in a MANET will be maintainedT09/8regularly by the underlying cluster formation algorithm.In a cluster,the heads rebroadcast can cover all other nodes in it’s cluster.To rebroadcast message to nodes in other clusters,gateway nodes are used,hence there is no need for a non-gateway nodes to rebroadcast the message.As different clusters may still have many gateway nodes,these gateways will still use any of the broadcasting approaches described in Section2to determine whether to rebroadcast or not. Ni et al[15]showed that the performance of the cluster based method where the location based approach was incorporated compared favourably to the orig-inal location based scheme.The method saved much more rebroadcasts and leads to shorter average broadcast latencies.Unfortunately,the reachability was unacceptable in low density MANETs.4Tree Based MethodsAlthough broadcasting using tree methods in wired networks is a well known and widely used technique,it is typically claimed to be inappropriate for MANETs because of their dynamic change in network topologies.On the contrary[25] has shown that the tree based method is an efficient,reliable and stable even in case of the ever changing network structure of the MANETs.The tree constructed in[25]was a spanning tree.The broadcasting using this tree is done by forwarding a broadcast message not to all neighbours but only to those who are neighbours in this tree.Since a tree is acyclic,each message is received only once by each node,giving advantages over the existing methods. Several proposed algorithms in the literature can be used for constructing and maintaining trees such as the spanning tree algorithm of bridged Ethernet net-works[26].However,most of these algorithms are developed to work in a stable networks and not in the constantly changing topology of a MANET.Authors in[27–31]are devoted to work in multicast trees and more specifi-cally[32]is dealing with multicast trees in MANETs.Their algorithms,however, are not suitable for handling the topology changes.Some algorithms although involve constructing a spanning tree,they are not appropriate for the mainte-nance of the tree in a dynamically changing topology.A feature that differentiates the tree based method reported in[25]from the other methods is that it specifically optimises the concept of one to one trans-missions.In this way the many drawbacks of local broadcasts do not affect the algorithm and thus,the method is very well suited to perform reliable broad-casts.Moreover,whilst all other methods require extra messaging to keep up network states,the proposed method in[25]has been designed to minimise this extra signalling traffic.More details on how the algorithm constructs and maintains the spanning tree,can be seen in[25].T09/95Tree Based Methods ImprovementThis section highlights some improvements that can be made to improve the tree based methods in Section4.In the context of one to one transmission mechanism(c.f.,[25]),it seems appropriate to propose an efficient and fast broadcasting scheme that aims at improving the broadcasting time in MANETs. Spanning trees with one to one transmission will not give an optimal time,bi-nomial trees are therefore suggested for this purpose.A binomial tree B k(c.f., Figure6)is an ordered tree defined recursively.For the binomial tree B k,there are n=2k nodes,the height of the tree is k and the root has degree k.The maximum degree of any node in a binomial tree is log2n.BkB0B k-1B k-1B B B B0123...B1B0B k-1B k-2B kB2B4Depth1234Figure6:A binomial tree B kA binomial tree structure is one of the most frequently used tree structures for parallel applications in various systems.Lo et al[33]has identified the bi-nomial tree as an ideal computation structure for parallel divide and conquer algorithms.Binomial trees can easily be embedded into a fully connected graphT09/10with constant dilation 1(c.f.,[34]).In a one to one transmission,binomial tree gives an optimal broadcasting time,this is because the informed number of nodes at most doubles at each step.For a binomial tree,there are exactly ni nodes at depth i for i =0,1,...,k ,andthe root has degree k ,which is greater than that of any other node,moreover all nodes are labelled as binary string of length k .Two nodes are connected if their corresponding strings differ precisely in one position.The root node (i.e.,a node at depth i =0)of the tree is labelled as 000...0where k binomial trees are attached.Nodes at depth i =1will have a single 1in their labels in ascending order from left to right (i.e.,u 1<u 2<u 3<...u (n i )).Nodes at depth i =2will have a two 1s in their labels in ascending order from left to right.This goes on to the last node of the tree at depth k with 111...1label.In the context of a one hop ad hoc network,the binomial tree will reduce thenumber of forwarding nodes into at most N 1(u )2−1as there N 1(u )2leaf nodes.The binomial tree can easily reconfigure as nodes leave and join the transmis-sion range.A node joining the transmission range will be connected to depth 1of the tree provided that |i |< n i where i =1and |i |represents the number of nodes at depth i .If |i |= n i for i =1,then the joining node will start a new level 0.If any node at any depth i,0≤i ≤k is leaving the transmission range r ,the far right node of depth 1(i.e.,u j ,1≤j ≤ n i )will take the position of the leaving node.No action will be taken if the far right node of depth 1is leaving or the root node with only one adjacent node to depth 1.For a multi hop MANET,the tree structure will be a heap of binomial trees.The MANET will look like a cluster of binomial trees.6Cluster Based Methods ImprovementsThe clustering method described in Section 3is so trivial and hence still suffers message redundancy and reachability problems.These problems can consider-ably be reduced by taking into account the MANET signal strengths.Nodes with strong signal strengths should be logically clustered together,this will en-hance nodes reachability and if tree based are incorporated,this approach will eliminate the message redundancy problem.By clustering the MANET nodes with strong signal strengths,the stability of the MANETs is guaranteed and hence tree based methods will ideally fit in this approach.There are numerous clustering methods in the literature,however,the majority of them are designed for wired networks.Clustering techniques usually demandT09/11intensive mathematical computations which will drain the MANETs battery power.7ConclusionsBroadcasting is an essential building block of any MANET,so it is imperative to utilize the most efficient broadcast methods possible to ensure a reliable network. This tutorial has offered an overview on all major broadcasting methods in the literature focusing on their functionalities and shortcomings and,moreover,sug-gesting improvements for some of them,as appropriate.Due to dynamic change of MANET topology and its scarce resource availability,however,there is no single optimal algorithms available for all relevant scenarios.References[1]V.Park and S.Corson.A Highly Adaptive Distributed Routing Algorithmfor Mobile Wireless COM97,1407-1415,1997.[2]Z.Haas.A New Routing Protocol for Reconfigurable Wireless Networks.ICUPC97,562-566,1997.[3]C.Perkins and E.Royer Ad hoc on Demand Distance Vector Routing.2nd IEEE Workshop on Mobile Computing Systems and Applications,3-12, 1999.[4]P.Sinha,R.Sivakumar and V.Bharghavan.CEDAR:A Core ExtractionDistributed Ad hoc Routing COM99,202-209,1999. [5]D.Johnson and D.Maltz.Dynamic Source Routing in Ad hoc WirelessNetworks.Mobile Computing,Academic Publishers,153-181,1996.[6]D.Johnson,D.Maltz and Y.Hu.The Dynamic Source Routing Protocol forMobile Ad hoc Networks.Internet Draft:draft-ietf-manet-dsr-09.txt,2003.[7]C.Perkins,E.Beldig-Royer and S.Das.Ad hoc on Demand Distance Vector(AODV)Routing.Request for Comments3561,July2003.[8]Z.Haas and M.Pearlman.The Performance of Query Control Schemesfor the Zone Routing Protocol.IEEE/ACM Transactions on Networking, 9(4):427–438,2001.[9]Z.Haas and B.Liang.Ad hoc mobility management with randomizeddatabase groups.Proceedings of the IEEE International Conference on Com-munications,1756-1762,1999.T09/12[10]Y.Ko and N.Vaidya.Location-aided Routing(LAR)in Mobile Ad hocNetworks.Proceedings of the ACM/IEEE International Conference on Mo-bile Computing and Networking(MOBICOM),66-75,1998.[11]mittee.Wireless LAN Medium ACCESS CONTROL(MAC)andPhysical Layer Specifications.IEEE802.11Standard.IEEE,New York, ISBN1559379359,1997.[12]B.Williams and parison of Broadcasting Techniques forMobile Ad hoc Networks.In Proceedings of the ACM Symposium on Mobile Ad Hoc Networking and Computing(MOBIHOC),194–205,2002.[13]C.Ho,K.Obraczka,G.Tsudik and K.Viswanath.Flooding for ReliableMulticast in Multi-hop Ad hoc Networks.International Workshop in Dis-crete Algorithms and Methods for Mobile Computing and Communication, 64–71,1999.[14]J.Jetcheva,Y.Hu,D.Maltz and D.Johnson.A Simple Protocol forMulticast and Broadcast in Mobile Ad hoc Networks.Internet Draft,draft-ietf-manet-simple-mbcast-01.txt,2001.[15]S.Ni,Y.Tseng,Y.Chen and J.Sheu.The Broadcast Storm Problem in aMobile Ad hoc Network.International Workshop on Mibile Computing and Networks,151–162,1999.[16]H.Lim and C.Kim.Multicast Tree Construction and Flooding in Wire-less Ad hoc Networks.In Proceedings of the ACM International Work-shop on Modeling,Analysis and Simulation of Wireless and Mobile Systems (MSWIM),2000.[17]W.Peng and X.Lu.Efficient Broadcast in Mobile Ad hoc Networks usingConnected Dominating Sets.Journal of Software,1999.[18]W.Peng and X.Lu.On the Reduction of Broadcast Redundancy in MobileAd hoc Networks.In Proceedings of MOBIHOC,2000.[19]W.Peng and X.Lu.AHBP:An Efficient Broadcast Protocol for MobileAd hoc Networks.Journal of Science and Technology,2002.[20]J.Sucec and I.Marsic.An Efficient Distributed Network-wide BroadcastAlgorithm for Mobile Ad hoc Networks.CAIP Technical Report248-Rut-gers University,September2000.[21]E.D.Kaplan.Understanding GPS:Principles and Applications.ArtechHouse,Boston,MA,1996.[22]M.Jiang,J.Li,and Y.C.Tay.Cluster Based Routing Protocol FunctionalSpecification.Internet Draft,1998.[23]M.Gerla and J.T.Tsai.Multicluster,Mobile,Multimedia Radio Network.ACM-Baltzer Journal of Wireless Networks,1(3):255–265,1995.T09/13[24]S.Alagar,S.Venkatesan,and J.Cleveland.Reliable Broadcast in Mo-bile Wireless Networks.In Military Communications Conference,MILCOM Conference Record,1:236-240,1995.[25]Alpr Jttner and dm Magi.Tree Based Broadcast in Ad hoc Networks.MONET Special Issue on WLAN Optimization at the MAC and Network Levels,2004.to appear.[26]IEEE Specification820.1d MAC Bridges D9,July14,1989.[27]A.Ballardie,P.Francis and J.Crowcroft.Core Based Trees(CBT)anArchitecture for Scalable Interdomain Multicast Routing.SIGCOMM93, San Francisco,85–95,1993.[28]D.Estrin.Protocol Independent MulticastSparse Mode(PIM-SM).Proto-col Specification RFC2362,June1998.[29]A.Adams,J.Nicholas and W.Siadak.Protocol Independent Multicast-Dense Mode(PIM-DM).Protocol Specification(Revised),IETF Internet-Draft,draft-ietf-pim-dm-new-v2-02.txt,October2002.[30]D.Waitzman,C.Partridge and S.Deering.Distance Vector MulticastRouting Protocol.RFC1075,November1988.[31]J.Moy.Multicast Routing Extensions for OSPF.CACM,37:61–66,1994.[32]C.Perkins,E.Royer and S.Das.Ad hoc On-demand Distance Vector(AODV)Routing.IETF Internet-Draft,draft-ietf-manet-aodv-11.txt,Aug 2002.[33]V.M.Lo,S.Rajopadhye,S.Gupta,D.Keldesn,M.A.Mohamed andJ.Telle.Mapping divide and conquer Algorithms to Parallel Architectures.Proc.of the1990International Conference on Parallel Processing.(March 1990),pp.128–135.[34]J.Wu and ardo and Y.Luo.Embedding of Binomial Trees inHypercubes with Link Faults.Journal of Parallel and Distributed Computing 10October(1998);54(1),pp.59–74.T09/14。