78

IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 11, NO. 1, FIRST QUARTER 2009

A Survey of Multicast Routing Protocols for Mobile Ad-Hoc Networks

Luo Junhai, Ye Danxia, Xue Liu, and Fan Mingyu
AbstractA Mobile Ad-hoc NETwork (MANET) is composed of Mobile Nodes (MNs) without any infrastructure. MNs selforganize to form a network over radio links. In this environment, multicast routing protocols are faced with the challenge of producing multi-hop routing under host mobility and bandwidth constraint. Multicast routing plays a signicant role in MANETs. In recent years, various multicast routing protocols with distinguishing feature have been newly proposed. In order to provide a comprehensive understanding of these multicast routing protocols designed for MANETs and pave the way for the further research, a survey of the multicast routing protocols is discussed in detail in this paper. Qualitatively, based on their primary multicast routing selection principle, we show that all these protocols could be placed under one of two broad routing selection categories: multicast routing based on application independence and multicast routing based on application dependence. Index TermsMobile ad-hoc network (MANET), mobile agent (MA), mobile node (MN), multicast routing protocol, quality of service (QoS), routing table (RT), survey.

MNs in MANETs are capable of communicating with each other without the use of a network infrastructure or any centralized administration. Due to the limited transmission range of wireless network interfaces, multiple hops may be needed for one node to exchange data with another across the network. In such a network, each MN operates not only as a host but also as a router, forwarding packets for other MNs in the network that may not be within direct wireless transmission range of each other. Each node participates in an ad-hoc routing protocol that allows itself to discover multi-hop paths through the network to any other node. As a promising network type in future mobile application, MANETs are increasingly attracting researchers [2], [3]. Multicast routing protocols belonging to different routing philosophies have been proposed in the literature. A proactive multicast routing protocol pre-determines the routes between any two nodes irrespective of the need for such routes. On the other hand, reactive multicast routing protocols discover routes only when required (i.e., on-demand). Some protocols consider all nodes are peers (at network topology), while others consider a hierarchy among nodes and only nodes in the same level of the hierarchy are treated as peers. Some protocols assume each node is aware of its current location in the network and also can learn the locations of other nodes in the network. Some multicast routing protocols that are sensitive to the available battery power at the nodes and the energy to be spent in packet transfer have been also proposed in the literature. Some multicast routing protocols discover and maintain multi-paths for a given node pair. The motivation and usage for these multiple paths depends on the protocols. This paper gives the state-of-the-art review of typical multicast routing protocols for MANETs. It is impossible to say which routing protocol is better for a given condition. Hence, the motivation is to group these multicast routing protocols under different routing strategies or categories and then compare these strategies. To our surprise, we nd that based on their primary routing selection principle, all of these protocols can be grouped under either application independent-based multicast routing or application dependent-based multicast routing strategies. Similarly, the results presented in this survey can be used by the research community and this can lead to a new paradigm for the comparison of multicast routing protocols [4]. Although there are already a few surveys in the area and some of them are even cited by this paper itself, some of them are out of date. This paper includes new technical trends such as overlay multicast, network coding-based multicast, energyefcient multicast etc. and the classication of the multicast

I. I NTRODUCTION ITH THE development of wireless communication technology, two basic wireless network models have been developed for the wireless communication system [1]. The xed backbone wireless model consists of a large number of Mobile Nodes (MNs) and relatively fewer, but more powerful, xed nodes. The communication between a xed node and a MN within its range occurs via the wireless medium. However, this requires a xed permanent infrastructure. Another system model, a Mobile Ad-hoc NETwork (MANET) [2], [3] , it is a self-organizing collection of MNs that form a temporary and dynamic wireless network on a shared wireless channel without the aid of a xed networking infrastructure or centralized administration. A communication session is achieved either through single-hop transmission if the recipient is within the transmission range of the source node, or by relaying through intermediate nodes otherwise. For this reason, MANETs are also called multi-hop packet radio network [4], [5]. However, the transmission range of each low-power node is limited to each others proximity, and out-of-range nodes are routed through intermediate nodes.

Manuscript received 30 November 2006; revised 24 November 2007. This work was partially funded by the National Natural Science Foundation of China under Grant No.60272091, the National Study-abroad Scholarship of China under Grant No. 27U38009, and the NSERC Discovery Fund. L. Junhai, Y. Danxia, and F. Minguy are with the School of Computer Science and Engineering, University of Electronic Science and Technology of China, Chengdu, 610054, China (luo.junhai2@mail.mcgill.ca). L. Junhai and X. Liu are with the School of Computer Science, McGill University, Montreal, H3A 2A7, Canada. Digital Object Identier 10.1109/SURV.2009.090107.

1553-877X/09/$25.00 c 2009 IEEE

JUNHAI et al.: A SURVEY OF MULTICAST ROUTING PROTOCOLS FOR MOBILE AD-HOC NETWORKS

79

protocols is a novel aspect of this article. We do not follow the classication methods of either the convention internet multicast or the methods of previous work, which already presented different survey studies in the area and provide enough insight on the classication of the current research work in the eld. Our primary goal is to provide a useful taxonomy of the eld of multicast routing protocol, which is comprehensive and up-to-the-minute. To accomplish this goal, we identify those basic components of a multicast routing protocol, break them down into the necessary separate mechanisms, and categorize properties we feel the mechanisms need to provide in order to fulll its function for the multicast routing protocol. The remainder of this paper is organized as follows. In Section II, we present a basic research on unicast and multicast routing protocols. Several characteristics of multicast routing protocols are given in Section III. In Section IV, we present the taxonomy of the multicast routing protocols and place them under either one of the two routing strategies. In Section V, we analyze some typical multicast routing protocols and describe the basic functionality of each of the multicast routing protocols. In Section VI, we conclude the survey by presenting a tabulated summary of the routing philosophies (e.g., proactive, reactive, at, hierarchical, location-awareness, power-sensitiveness etc), the routing metrics and the primary routing selection principles. Finally, we draw conclusions and discuss directions of future work about the multicast routing protocols for MANETs. II. ROUTING P ROTOCOLS FOR MANET S A. Unicast Routing Protocols Routing is the most fundamental component in networks to support data communications. To make MANETs practical, efcient and effective unicast routing protocol is being a critical issue. Many different unicast routing protocols [4] have been developed for MANETs. They can be classied into two types of unicast routing methodologies as follows: A proactive unicast routing protocol is also called a tabledriven unicast routing protocol. Using the proactive unicast routing protocol, nodes continuously evaluate routes to all reachable nodes and attempt to maintain consistent, up-to-date routing information. Therefore, the source node can get a routing path immediately if it needs one. In the proactive unicast routing protocols, all nodes need to maintain a consistent view of the network topology. When a network topology change occurs, respective updates must be propagated throughout the network to notify the change. Most proactive unicast routing protocols proposed have inherited properties from algorithms used in wired networks. To adapt to the dynamic features of MANETs, necessary modications have been made on traditional wired network unicast routing protocols. Using the proactive unicast routing algorithms, MNs proactively update network state and maintain a route regardless of whether the data trafc exists or not, the overhead to maintain up-to-date network topology information is high. The Optimized Link State Routing protocol (OLSR) [6] and Dynamic destinationSequenced Distance-Vector protocol (DSDV) [7] are examples for proactive routing protocols for MANETs. A different approach from the proactive unicast routing is the reactive unicast routing. The reactive routing protocol

is also called source-initiated "on-demand" unicast routing protocol. This type of unicast routing creates routing only when desired by the source node. When a node requires a routing to a destination, it initiates a routing discovery process within the network. This process is completed once a route is found or all possible routing permutations have been examined. Active routes may be disconnected due to node mobility in MANETs. Therefore, route maintenance is an important operation of reactive routing protocols. The Dynamic Source Routing protocol (DSR) [8] and Ad-hoc Ondemand Distance Vector protocol (AODV) [9] are examples for reactive routing protocols. Compared to the proactive routing protocols, less control overhead is a distinct advantage of the reactive routing protocols. Thus, reactive routing protocols have better scalability than proactive routing protocols. However, when using reactive routing protocols, source nodes may suffer from long delays for route searching before they can forward data packets. B. Multicast Routing Protocols Applications of MANETs are in areas where rapid deployment and dynamic reconguration are necessary, but the wired network is not available. These include military battleelds, emergency search, rescue sites, classrooms, and conventions where participants share information dynamically using their mobile devices. These applications lend themselves well to multicast operation. In addition, within a wireless medium, it is even more crucial to reduce the transmission overhead and power consumption. Multicasting can be used to improve the efciency of the wireless link when sending multiple copies of messages to exploit the inherent broadcast property of wireless transmission. So multicasting plays an important role in MANETs. In the wired environment, there are two popular network multicast approaches, namely, the shortest path multicast tree and core-based tree. The shortest path multicast tree guarantees the shortest path to each destination. But each source needs to build a tree. Usually, there exist too many trees in the network, so the overhead tend to be large. In contrast, the core-based tree constructs only one tree for each group and the number of trees is greatly reduced. Unlike typical wired multicast routing protocols, multicast routing for MANETs must address a diverse range of issues due to the characteristics, such as, low bandwidth, mobility and low power. MANETs delivers lower bandwidth than wired networks. Therefore, the information collection is expensive during the formation of a Routing Table (RT). Mobility of nodes, which causes topological changes of the underlying network, also increases the volatility of network information. In addition, the limitation of power often leads users to disconnect mobile units. Multicast routing protocols have emerged as one of the most active research areas. There are three basic categories of multicast methods in MANETs [4]. (1) A basic method is to simply ood the network. Every node receiving a message oods it to a list of neighbors. Flooding a network acts like a chain reaction that can result in exponential growth. (2) The proactive approach pre-computes paths to all possible

80

IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 11, NO. 1, FIRST QUARTER 2009

destinations and stores this in formation in the RT. To maintain an up-to-date database, routing information is periodically distributed throughout the network. (3) The nal method is to create paths to other nodes on demand. The idea is based on a query response mechanism or reactive multicast. In the query phase, a node explores the environment. Once the query reaches the destination the response phase starts and establishes the path. Recently, many multicast routing protocols have been newly proposed. The Ad-hoc Multicast Routing protocol utilizing increasing Id numberS (AMRIS) [10] builds a shared-tree to deliver multicast data. Each node in the multicast session is assigned an ID number and it adapts to connectivity changes by utilizing the ID numbers. The Multicast Ad-hoc On-Demand Vector (MAODV) [11] stems from the use of a destination sequence number of each multicast entry. The sequence number is generated by the multicast group-head to prevent loops and to discard stale routes. The Ad-hoc Multicast Routing (AMRoute) [12] is also a shared-tree protocol which allows dynamic core migration based on group membership and network conguration. The Lightweight Adaptive Multicast (LAM) algorithm [13] is a group shared-tree protocol that does not require timer-based messaging. Similar to the OnDemand Multicast Routing Protocol (ODMRP) [14], the Core Assisted Mesh Protocol (CAMP) [15] uses a mesh. However, a conventional routing infrastructure based on enhanced distance vector algorithm or link state algorithm is required for CAMP to operate. Core nodes are sued to limit the trafc required when a node joins a multicast group. The Location Guided Tree (LGT) [16] is a small group multicast scheme based on packet encapsulation. It builds an overlay multicast packet distribution tree on top of the underlying unicast routing protocol. The Differential Destination Multicast (DDM) [17] can be viewed as ooding with limited scope, wherein the ooding is contained within selected Forwarding Group (FG) nodes. Most of these multicast routing protocols are primarily based on avors of distance-vector or link-state routing plus additional functionalities to assist the routing operations in particular ways. The goals of all these protocols include such as minimal control overhead, minimal processing overhead, multi-hop routing capability, dynamic topology maintenance or loop prevention. However, many multicast routing protocols do not perform well in MANETs because in a highly dynamic environment, nodes move arbitrarily, thus network topology changes frequently and unpredictably. Moreover, bandwidth and battery power are limited. These constraints in combination with the dynamic network topology make multicast extremely challenging. III. C HARACTERISTICS OF M ULTICAST ROUTING P ROTOCOLS Characteristics of multicast routing protocols are very important for researchers and designers to help to understand a multicast routing protocol and nd its relationship with others. These characteristics mainly are related to the information, which is exploited for MANETs, when this information is acquired, and the roles which nodes may take in the multicast routing process.

Location services for MANET

Flooding-based

Quorum-based

Reactive

Proactive

Explicit-quorum

Implicit-quorum

Hierarchical

Flat

Hierarchical

Flat

Hierarchical

Flat

Flat

Hierarchical

Fig. 1.

Location services for multicast routing protocols.

A. Evaluation Principles for Multicast Routing Protocols Most of the multicast routing protocols assume physically at network architecture with MNs having homogeneous capability in terms of network resources and computing power. In practice, however, this assumption often may not hold true since there exist various types of MNs with different role, capacity and mobility pattern. In architecture-based multicast routing protocols, MANETs have physically hierarchical architecture, where different types of MNs form an ad-hoc network hierarchy. For example, the Hierarchical QoS Multicast Routing Protocol (HQMRP) [18] builds a multicast structure at each level of the hierarchy for efcient and scalable multicast message delivery. And the SelfOrganizing Map (SOM) [19] is also a typical hierarchical architecture, which provides a way for automatically organizing the hierarchical architecture. In location-based multicast routing protocols, the availability of a Global Positioning System (GPS), Bluetooth or other locations systems easily gets geographical information of MNs when needed [20]. Each node determines its own location through the use of GPS or some other type of positioning service. A location service is used by the sender of a packet to determine the location of the destination. The routing decision at each forwarding node is then based on the locations of the forwarding nodes neighbors and the destination node. Fig. 1 describes the classication of location services for locationbased multicast routing protocols [21]. The Location-based Geocasting and Forwarding (LGF) [22], LGT and the Scalable Position-Based Multicast (SPBM) [23] protocol are typical location-based multicast routing protocols. At the top level of the taxonomy, location services for multicast protocols can be divided into ooding-based and quorum-based approaches. Flooding-based protocols can be further divided into proactive and reactive approaches. In the proactive ooding-based approach, each (destination) node periodically oods its location to other nodes in the network, each of which maintains a location table recording the most recent locations of other nodes. The interval and range of such ooding can be optimized according to the nodes mobility and the distance effect. In reactive (on-demand) ooding-based approaches, if a node can not nd a recent location of a destination to which it is trying to send data packets, it oods a scoped query in the network in search of the destination. In quorum-based protocols, all nodes (potential senders or receivers) in the network agree, implicitly or explicitly, upon a mapping that associates with each nodes unique identier to one or more other nodes in the network. In the explicit quorum-based approach, each location update of a node is

JUNHAI et al.: A SURVEY OF MULTICAST ROUTING PROTOCOLS FOR MOBILE AD-HOC NETWORKS

81

MANET multicast routing protocols

Application independence

Application dependence

Topology -based

Initialization -based

Maintenance -mechanism

QoS

Energy -efficiency

Reliable

Networkcoding

Delay

Energy-saving

Absolute

Bandwidth

Destination-initiation

Hard-state

Tree-based

Source-initiation

Receiver-centered

Cost-optimization

Power-aware

Mesh-based

Soft-state

Hybrid

Maximum-throughput

Fig. 2.

Taxonomy of multicast routing protocols.

sent to an explicit dened subset (update quorum) of available nodes, and a location query for that node is sent to a potentially different subset (query quorum). In the implicit quorum-based protocols, location servers are chosen via a hashing function, either in the node identier space or in the location space. Notice that the implicit quorum-based protocols are sometimes called rendezvous hashing-based protocols. B. Performance Criteria for Multicast Routing Protocols Although numerous multicast routing protocols have been proposed for MANETs, there is no a one-for-all scheme that works well in scenarios with different network sizes, trafc overloads and node mobility patterns. Moreover, those protocols are based on different design philosophies to meet specic requirements of different application domains. Thus, the performance of a multicast routing protocol may vary dramatically with the variations of network status and trafc overhead. It is a very difcult make to give a comprehensive performance comparison for a large number of multicast routing protocols. There are three different ways to evaluate and compare the performances of multicast routing protocols as follows: 1) User parameters and congurations; such as average multicast degree, control overhead, average delay, throughput and multicast service cost; 2) Different updating methods; Multicast routing updating can be done in one of three ways: 1) Store and update: store the information in a RT and update it by listening to routing messages; 2) Delete all and refresh: discard all old routes (timeout) and start over again; 3) Unicast protocol support: use the services of a separate unicast routing protocol for routing updating; 3) The performance is evaluated by different simulation tools, such as NS-2, Opnet, Matlab, CASSAP, GloMoSim and SPW. With the popularity of MANETs and considering the dynamic network features, integrated criteria, such as philosophies of Quality of Service (QoS), power control, reliable

and secure, for evaluating performances of multicast routing protocols should be proposed to meet the different mobile application requirements in different environments and different design targets. IV. TAXONOMY OF M ULTICAST ROUTING P ROTOCOLS The goal of MANETs is to extend mobility or wired network into the realm of autonomous, mobile, wireless domains, where a set of nodes forms the network routing infrastructure in an ad-hoc fashion. MANETs extends applications of the wired network into some special applications domains that result in designing new multicast routing protocols to meet those special applications in MANETs. Qualitatively, based on their primary selection routing principles, we show that all these protocols can be placed under one of two broad routing selection categories: multicast routing protocols based on application independence and application dependence. That is, inherent in the way a description is written is the purpose for which the application will use the information. The taxonomy of the multicast routing protocols shown in Fig. 2 is studied in this paper. A. Application Independence-Based Multicast Routing Protocols Multicast Routing protocols based on application independence stem from the conditional multicast routing protocols in the wired environment, where there are two popular networks multicast schemes: the shortest path multicast tree and corebased tree. Most multicast routing protocols based on application independence for MANETs use "hop-number" as a metric. If there are multiple routing paths available, the path with the minimum hop number will be selected. If all wireless links in the network have the same failure probability, short routing paths are more stable than the long ones and can obviously decrease trafc overhead and reduce packet collisions. The application independence-based multicast routing protocols is to maintain the state of a link considered in the routing construction phase. And existing multicast routing approaches

82

IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 11, NO. 1, FIRST QUARTER 2009

Topology based multicast routing protocols

Hybrid
Fig. 3.

Tree-based

Mesh-based

Topology-based multicast routing protocols.

for MANETs can be divided into topology, initialization and maintenance-mechanism multicast routing protocols. 1) Topology-Based Multicast Routing Protocols: One of the most popular methods to distinguish MANETs multicast protocols is based on how distribution paths among group members are constructed. In terms of this method, existing multicast routing protocols for MANETs can be divided into tree-based, mesh-based and hybrid-based multicast routing protocols (see Fig. 3). Tree-based multicast routing protocols can be further divided into source-rooted and core-rooted schemes according to the roots of the multicast trees. In a source-rooted tree-based multicast routing protocol, source nodes are roots of multicast trees and execute algorithm for distribution tree construction and maintenance. This requires that a source must know the topology information and addresses of all its receivers in the multicast group. Therefore, the source-rooted tree protocols suffer from control trafc overhead when used for dynamic networks. AMRoute is an example for the source-rooted tree multicast routing protocol for MANETs. In a core-rooted tree multicast routing protocol, cores are nodes with special functions such as multicast data distribution and membership management. Some core-rooted multicast routing protocols utilize tree structures also, but unlike source-rooted tree-based multicast routing, multicast trees are rooted at core nodes. For different source-rooted multicast routing protocols, core nodes may perform various routing and management functions. The Shared Tree Ad-hoc Multicast Protocol (STAMP) [24] and the Adaptive Corebased Multicast Routing Protocol (ACMP) [25] are core-based multicast routing protocols proposed for MANETs. The tree-based protocols establish a single path between any two nodes in the multicast group. These protocols require a minimum number of copies per packet to be sent along the branches of the tree. Therefore, they are bandwidth efcient. If there is only one source, only a minimal number of nodes are involved in the routing. Hence, tree-based approaches could also be relatively power efcient. However, as mobility increases, link failures trigger the reconguration of the entire tree. Either has to maintain a shared tree, losing path optimality, or maintain multiple trees resulting in storage and control overhead when there are many sources. In a mesh-based multicast routing protocol, packets are distributed along mesh structures that are a set of interconnected nodes. Routing discovery and mesh building are accomplished in two ways: by using broadcasting to discover routes or by using core or central points for mesh building. The meshedbased protocols have high robustness in comparison with the tree-based protocols in the high mobility environment as they provide redundant paths from source to destinations

while forwarding data packets. To maintain the mesh topology however, it requires more control messages than the treebased approach since multiple copies of the same packet are disseminated through the mesh, resulting in power inefciency, network load and control overhead. The Mesh-based Multicast Routing Protocol with Consolidated Query Packets (CQMP) [26], the Enhanced On-Demand Multicast Routing Protocol (E-ODMRP) [27] and the Bandwidth Optimized and Delay Sensitive (BODS) [28] are mesh-based multicast routing protocols proposed for MANETs. Hybrid-based multicast routing protocols combine with the advantages of both tree and meshed-based approaches. Hence, hybrid protocols address both efciency and robustness. Using this scheme, it is possible to get multiple routing paths, and duplicate messages can reach a receiver through different paths. However, they may create non-optimal trees with nodes mobility. The Efcient Hybrid Multicast Routing Protocol (EHMRP) [29] is an instance for hybrid-based multicast routing protocol. 2) Initialization-Based Multicast Routing Protocols: Multicast initialization is the mechanism by which a multicast session is created and advertised to nodes within the network. Nodes that are interested in joining the multicast session (herein known as I-Nodes) join in the initialization phase. Nodes that are not interested in joining the multicast session are herein known as U-Nodes. We will describe these two types of initialization as follows: 1) In the sender initialization routing protocols, each packet sent by a sender is acknowledged by each receiver. The sender takes the responsibility for data delivery by maintaining the state information of each receiver and processing the feedback from receivers. 2) In the receiver initialization routing protocols, its the responsibility for detecting transmission error and packet loss. A receiver detects a packet loss when it nds a gap in sequence number of received packets. 3) Maintenance-Mechanism-Based Multicast Routing Protocols: There are two different possible styles for maintenance-mechanism-based multicast routing protocols, namely, the hard state (HS) protocol (also called connection-oriented), and the soft state (SS) protocol (also called connectionless). 1) Under the HS protocol, in the absence of some event to trigger a protocol response, the HS protocols state will remain unchanged or hard for an unbounded time period. This status is created and deleted in a fully deterministic manner, so the HS setup protocol must be reliable with acknowledgments and retransmissions. 2) Under the SS protocol, it regards the state as cached information that is installed and periodically refreshed by the end hosts. If the routing changes, the refresh messages automatically install the necessary state along the new routing. The design and exibility allow its operation to be biased towards high reactivity (i.e., low time complexity) and bandwidth conservation (i.e., low communication complexity) rather than routing optimality making it potentially well-suited for use in dynamic wireless networks.

JUNHAI et al.: A SURVEY OF MULTICAST ROUTING PROTOCOLS FOR MOBILE AD-HOC NETWORKS

83

B. Application Dependence-Based Multicast Routing Protocols Compared with the wired network applications, some special applications in MANETs introduce some new problems in addition to the ones already present in xed networks. Multicast routing protocols based on application dependence are designed for those special applications when needed. And application dependence-based multicast routing protocols can further divided into QoS, energy-efciency network-coding and reliable multicast routing protocols as follows: 1) QoS-Based Multicast Routing Protocols: Most of conventional multicast routing protocols are designed for minimizing data trafc or minimizing the average hops number for delivery a packet in the network. When Qos is considered, some protocols may be unsatisfactory or impractical due to the lack of resource, the excessive computation overhead, the lack of knowledge about the global network state or the excessive message processing overhead. Even, some multicast routing protocols, such as LGT, AMRIS and CAMP are designed without explicitly considering QoS. QoS multicast routing requires not only to nd a routing from a source to a destination, but to satisfy the end-to-end QoS requirement, often given in terms of bandwidth or delay. QoS is more difcult to guarantee in MANETs than in other type of networks, because the wireless bandwidth is shared among adjacent nodes and the network topology changes as the nodes move. This requires extensive collaboration between the nodes, both to establish the routes and to secure the resources necessary to provide the QoS. With the extensive applications of MANETs in many domains, the appropriate QoS metrics should be used, such as bandwidth, delay, packet loss rate and cost for multicast routing. Therefore, QoS multicasting routing protocols face the challenge of delivering data to destinations through multihop routes in the presence of node movements and topology changes. The QoS Multicast Routing Protocol for Clustering mobile Ad-hoc network (QMRPCAH) [30] is an example for QoS multicast routing protocols for MANETs. Fig. 4 shows the classication of QoS multicast routing protocols for MANETs. Interaction-based between the routing protocol and the QoS provisioning mechanism can be classied into two categories, coupled and decoupled QoS protocol. In the case of the coupled QoS protocol, the routing protocol and the QoS provisioning mechanism closely interact with each other for delivering QoS guarantees. If the routing protocol changes, it might fail to ensure QoS guarantees. But in the case of decoupled protocol, the QoS provisioning mechanism does not depend on any specic routing protocol to ensure QoS guarantees. Similarly, interaction-based between the routing protocol and the MAC protocol, QoS protocols can be classied into two categories, independent and dependent QoS protocols. In the independent QoS protocol, the network layer is not dependent on the MAC layer for QoS provisioning. The dependent QoS protocol requires the MAC layer to assist the routing protocol for QoS provisioning. Finally, routing information updating mechanism-based, QoS protocols can be classied into three categories viz., table-driven, on-demand, and hybrid QoS protocols. In the table-driven protocol, each node in the network maintains a RT which aids to forwarding
Interaction based between routing protocol and QoS provisioning mechanism

QoS multicast routing protocols

Interaction based between network and MAC layer

Routing information updating mechanism

Coupled

Decoupled

Independent

Dependent

On-demand

Table-driven

Hybrid

Fig. 4.

Classication of QoS multicast routing protocols.

packets. In the on-demand protocol, no such tables are maintained the nodes, so the source node has to discover the route on the y. The hybrid protocol incorporates features of both the table-driven and the on-demand protocols. 2) Energy-Efciency Multicast Routing Protocols: Any form of infrastructure and nodes in MANETs are typically powered by batteries with a limited energy supply, each node ceases to its function when it exhausts its available energy. This suspension may potentially result in partitioning of the entire network. Therefore, given the energy constraints placed on the network nodes, designing energy-efcient multicast routing protocols is an important issue for MANETs, maximizing the lifetime of its nodes and thus of the network itself. Such protocols either start with an empty solution which is gradually augmented to a multicast tree (augmentation algorithms) or take as input an initial multicast tree and walk on different multicast trees for a nite number of steps until some acceptable decrease in energy consumption is achieved (local search algorithms) [31]. 1) Augmentation protocols build a multicast tree by starting from an empty solution which is gradually augmented until a guest network having directed paths from the root to the terminals is established. Clearly, once such a guest network is available, it can be easily converted to a multicast tree. The solution is augmented in phases. In each phase, an augmentation algorithm is added to the solution scheme, such as an edge, a path and a spider. The structure is selected among all candidate structures so that a local objective is minimized. The local objective is usually related to the energy needed in order to establish the edges of the structure. 2) Local search protocols perform a walk on multicast trees. The walk starts from a multicast tree given as input. In each step, a local search algorithm moves to a new multicast tree obtained by removing some of the edges of the previous one and adding new edges, so that the necessary connectivity properties are maintained. The rule used in each move for selecting the next multicast tree is related to energy. Since local search algorithms require a multicast tree to start walking on, they are usually called after an augmentation algorithm. The Minimum Weight Incremental Arborescence (MWIA) [32] and the Power-Controlled Hybrid Multicast Routing protocol (PCHMR) [33] are examples for energy-efciency multicast routing protocols. Energy-efciency[31] multicast protocols is also classied into ve categories: active energy saving protocols, maximizing network lifetime protocols, passive energy saving

84

IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 11, NO. 1, FIRST QUARTER 2009

Energy-efficiency multicast routing protocols

Activity-based

Connectivity-based

Multicasting

Unicasting

Topology control

Passive energy saving

Active energy saving

Maximizing network lifetime

Fig. 5.

Classication of energy-efciency multicast routing protocols.

protocols, topology control protocols, and energy-efciency multicasting protocols (see Fig. 5). In active energy saving protocols, the main goal of those routing protocols is nding a routing path with minimum energy consumption. The intention is to minimize the energy consumed per packet. The primary focus of maximizing network lifetime protocols is to consume node energy in a more balanced manner. To consume node energy in a more balanced manner, an intuitive technique is to utilize a cost function based on the nodes remaining battery capacity (i.e., the energy left in the battery). General goal of the passive energy saving protocols in this category is to turn off as many radios as possible while still maintaining the necessary network connectivity. In topology control protocols, the network topology is formed by the links of each node and the number of links a node has is mainly determined by its transmission power. By managing its transmission power wisely, a node is able to not only maintain all necessary links, but to reduce its power consumption. The energy-efciency multicast protocol is an extension of Broadcast Incremental Power (BIP). 3) Network-Coding Multicast Routing Protocols: Network coding [34], [35] has been proved to be an effective way to achieve the maximum ow capacity in multicast network, which is bounded by the famous maximum-ow minimumcut theorem. This technique is specially t to some new types of network, e.g., ad-hoc, sensor network etc. Given a network with capacity constraints on links, one problems of designing multicast routing protocols is to compute the maximum multicast throughput possible for communication between a source node and a set of receivers. The main advantage of using network coding can be seen in multicast scenarios. Network coding enables better resource utilization and can achieve the max-ow which is the theoretical upper bound of network resource utilization through a network node, such as a router to encode its received data before forwarding it. Each node implements with network coding function, receives information from all the input links, encodes it and sends the encoded information to all output links. The coded network lends itself, for multicast connections, to a cost optimization which not only outperforms traditional routing tree-based approaches, but also lends itself to a distributed implementation and to a dynamic implementation when changing conditions, such as mobility, arise. It [36] shows that under a simplied layered model, the minimum energy multicast problem is solvable as a linear program, assuming network coding.

4) Reliable Multicast Routing Protocols: In ad-hoc network environment, every link is wireless and every node is mobile. Those features cause increased loss easily, unreliable as well as multicasting inefcient. It is a major challenge to transmission delays and packet losses due to link changes of a multicast tree at the provision of high delivery ratio for each packet transmission. Reliable multicast routing protocol becomes a very challenging research problem for MANETs. The design of reliable multicast depends on the following three decisions: (1) By whom errors are detected; (2) How error messages are signaled and (3) How missing packets are retransmitted. These protocols have different design principles and operational features in addressing the reliability issue. Other protocols opt for another set of properties while some protocols favor one set of features. The reliable multicast protocols can be classied into the following four types: Sender-initiated, where all receivers send ACKs for each packet that they receive; Receiver-initiated, where the receivers send NACKs on detection of transmission error or packet loss; Ring-based, where receivers are organized in a logical ring and receivers take turns to acknowledge the packets received to ensure reliability; and Tree-based, where receivers are organized into subgroups to relieve the sender by processing all control messages from all receivers. The Reliable Adaptive Multicast Protocol (RAMP) [37] and the Reliable On-demand Routing Protocol (RORP) [38] are examples for reliable multicast routing protocols. V. S URVEY OF T YPICAL M ULTICAST ROUTING P ROTOCOLS Because many multicast routing protocols have been proposed for MANETs, it is impossible to cover all of them in this review. In this section, the multicast routing protocols are discussed in detail and distinct features, inheriting relationships and performance characteristics of these routing protocols can be evaluated. There are some selection criteria for the multicast routing protocols in the paper as follows: 1) Protocols which are just popular and reect the state-ofthe-art of research work on multicast routing protocols. 2) Protocols which cover the major progress on this specic topic and inspire other researchers on which potential directions they should work. 3) Protocols which present new ideas, new technical trends and are currently in practical use. 4) Protocols which are simple and easy to understand through sufciently introducing the fundamental concepts and background of multicast. 5) Protocols which are published in top international conferences or journals. 6) Protocols which are abbreviated and easy to be remembered for the researchers in this led. A. The Shared Tree Ad-hoc Multicast Protocol (STAMP) The Shared Tree Ad-hoc Multicast Protocol (STAMP) [24] is a reactive core-rooted multicast routing protocol for MANETs, which is independent from the underlying unicast routing protocol in order to achieve efcient and adaptive multicast communications rstly inside each cluster and secondly

JUNHAI et al.: A SURVEY OF MULTICAST ROUTING PROTOCOLS FOR MOBILE AD-HOC NETWORKS

85

among the clusters. In STAMP, a source of a multicast group does not need to join the multicast delivery structure to send a datagram to the group. Multicast datagram is sent on the shortest paths between the sources and the core. As soon as a data packet reaches a tree member, it is forwarded on the tree. Finally, a distributed mechanism is used to elect the core node among the receiver nodes of a specic multicast group. Therefore, unlike CAMP operation, core nodes are not pre-assigned. STAMP combines the advantages of both meshbased and tree-based protocols and achieves high delivery ratio even under heavy mobility and heavy trafc. B. The Adaptive Core-Based Multicast Protocol (ACMP) The Adaptive Core-based Multicast routing Protocol (ACMP) [25] is an on-demand, source-oriented group-shared tree multicast routing protocol. ACMP is trying to nd tradeoff between routing overhead and data transmission efciency. It uses a tree structure to connect all group members on demand. A core is the rst source of a multicast session for group members to join the multicast group. If no core exists in the network, it is not necessary to construct and maintain tree and all receivers would remain silent. ACMP selects the core to give the indication of multicast data so that a multicast structure can be constructed and maintained only when there are requirements. Core also limits the control trafc for group members to join the multicast group. Routing message takes a little percentage of trafc in the network, so data transmission efciency is more important. To improve the performance of tree structure, ACMP detects link failure during data forwarding, and uses two phases, local route recovery and periodical Mtree refreshing, to maintain an optimal multicast tree. And it is economic in bandwidth and power consumption since it sends a few packets for delivering a data packet to receivers. C. The Mesh-Based Multicast Routing Protocol With Consolidated Query packets (CQMP) The Mesh-based multicast routing Protocol with Consolidated Query packets (CQMP) [26] is a reactive mesh-based multicast routing protocol with an idea of query packet consolidation to address this scalability problem. It retains all of the advantages of the ODMRP, such as high packet delivery ratio under high mobility, high throughput. Moreover, the protocol signicantly reduces control overhead, one of the main weaknesses of ODMRP, under the presence of multiple sources. This feature is a crucial contributing factor to the scalability of multicast routing for MANETs. Instead of each source sending advertising packets to the network, in CQMP, each core disseminates to the network the mappings of multicast addresses to one or more core addresses. CQMP, however, assumes the availability of routing information from a unicast routing protocol. This unicast routing protocol is also required to provide correct distances to known destinations within a nite amount of time. CQMP also assumes the existence of a beaconing protocol, which may be embedded into the unicast routing protocol. In addition, CQMP relies on the associated routing protocol to work correctly in the presence of router failures and network partitions.

D. The Enhanced On Demand Multicast Routing Protocol (EODMRP) The Enhanced On-Demand Multicast Routing Protocol (EODMRP) [27] is an enhancement of ODMRP, which is a reactive mesh-based multicast routing protocol. It is an enhanced version of ODMRP with adaptive refresh. Adaptation is driven by receivers reports. The second enhancement is the unied local recovery and receiver joining scheme. As the time between refresh episodes can be quite long, a new node or a momentarily detached node might loose some data while waiting for the routing to it to be refreshed and reconstructed. Upon joining or upon detection of broken route, a node performs an expanding ring search to proactively attach itself to forwarding mesh or to requests a global route refresh from the source. Compared to ODMRP, a slightly lower packet delivery ratio might be expected in E-ODMRP in light load since the new scheme uses packet loss as indicator of a broken link. The major advantage is reduced overhead, which translates into a better delivery rate at high loads, yet keeping the same packet delivery ratio as the original ODMRP. E. The Bandwidth Optimized and Delay Sensitive protocol (BODS) The Bandwidth Optimized and Delay Sensitive (BODS) [28] is a source-rooted mesh multicast routing protocol in a distributed manner. It constructs a multicast delivery structure based on nearest participant heuristic, which is more optimal bandwidth-optimal multicast delivery structure in terms of bandwidth consumption without sacricing delay performance. The effectiveness of this algorithm is veried by integrating BODS into ODMRP protocol. BODS can achieve similar or better packet delivery ratio with a reduction of data overhead and improve the delay performance of the network especially under high trafc load. This is particularly important for bandwidth-avid and delay-sensitive applications such as multimedia streaming in a bandwidth-limited MANETs. Moreover, being a multicast path setup protocol, BODS is a general protocol that can be integrated into any existing mesh-based multicast routing protocols. A possible extension is to analyze the performance of BODS from a theoretical perspective. Competitive analysis which is commonly used in the analysis of centralized online algorithm might be extended to evaluate the performance of BODS. F. The Efcient Hybrid Multicast Routing Protocol (EHMRP) The Efcient Hybrid Multicast Routing Protocol (EHMRP) [29] is a hybrid multicast routing protocol to be suitable for high mobility applications and improve the scalability of the ODMRP. It separates out data forwarding path from joinquery forwarding path. EHMRP incorporates low overhead local clustering technique to classify all nodes into core and normal categories. When multicast routes to destination nodes are unavailable, join-query messages are sent to all nodes in the network and data packets are forwarded by the core nodes to the destination nodes using DDM, which is a stateless multicast approach where multicast tree information is appended with each data packet header. EHMRP does not require any underlying unicast protocol. There are key components of

86

IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 11, NO. 1, FIRST QUARTER 2009

EHMRP as follows: (1) Classifying core and normal nodes; (2) Separating out data forwarding path while sending join request and sending data packets using DDM; (3) Separate handling of received data packets coming through DDM path; (4) Group membership update; (5) Normal functionality of ODMRP protocol. G. The RObust Multicasting in Ad-Hoc Network Using Tree (ROMANT) The RObust Multicasting in Ad-hoc Network using Tree (ROMANT) [39] is a reactive tree-based multicast routing protocol. Instead of using a new kind of control packet, the existing control packet, the group hello is used to avoid the problem in xing broken links faced by MAODV. ROMANT xes the performance problems faced by MAODV (high control overhead and low packet delivery ratio in situations of high mobility, high trafc load and a large number of members). Moreover, ROMANT does not introduce new problems. The process of merging of partitions in ROMANT is much simpler than that of MAODV. ROMANT eliminates the drawbacks of MAODV and avoids any dependency on unicast routing protocols without incurring any extra overhead. It also provides equal or better packet delivery ratio than ODMRP at only a fraction of the total overhead incurred by ODMRP. H. The Optimized Polymorphic Hybrid Multicast Routing Protocol (OPHMR) The Optimized Polymorphic Hybrid Multicast Routing protocol (OPHMR) [41] is a proactive, polymorphic energy efcient and hybrid multicast routing protocol. It attempts to benet from the high efciency of proactive behavior and the limited network trafc overhead of the reactive behavior, while being power, mobility, and vicinity-density aware. The protocol is based on the principle of adaptability and multi-behavioral modes of operations. It is able to change behavior in different situations in order to improve certain metrics like maximizing battery life, reducing communication delays, improving deliverability, etc. OPHMR denes four different behavioral modes of operation, two power level thresholds, one mobility level threshold and one vicinity density thresholds. Under the four different modes, the lifetime of its corresponding entry is also different. Power threshold determines the nodes behavior in order to extend its battery life. Speed threshold is required to maintain better connectivity and awareness of the topology changes. Density threshold is considered when the mobility speed is high. I. The Mobile Agents Aided Multicast Routing Protocol (MAMR) The Mobile Agents aided Multicast Routing protocol (MAMR) [42] is a reactive QoS-based hybrid multicast routing protocol where intelligent MAs can be used with any ondemand multicast routing protocol. MAMR can integrate with other exiting multicast routing protocols, such as MAODV and ODMRP in order to overcome the limitation that most of multicast routing protocols try to discover the routing on demand by ooding route request messages. In MAMR, MAs

are simple packets, which move over the network and provide the current topology information and other QoS values such as link delay, congestion etc, which helps nodes for taking efcient routing decisions as they visit different nodes. The information carried by the MAs helps to nd a route for a given destination, when no route exists in the multicast table to the destination. By this way, the protocol overcomes the additional delay which would have been required, in nding a new route to the destination and also reduces the control trafc generated. And the availability of this route information at nodes will avoid routing protocols in doing broadcast route discovery and hence reduces end-to-end latency of the network. Although this method requires extra cost for processing MAs, the benets would be gained in terms of better end-toend latency and packet delivery ratio. J. The Multicast Power Greedy Clustering protocol (MPGC) The Multicast Power Greedy Clustering protocol (MPGC) [43] is an adaptive power-aware and on-demand multicasting protocol with the mesh scheme. It rst forms the hierarchical cluster structures with greedy power control where each node can adjust exibly its transmission power to t individual geographical location. MPGC uses greedy heuristic clustering, power aware multicasting and clustering maintenance that try to be energy efcient and prolong the network lifetime. MPGC assumes that each node has multiple power levels for transmission and any cluster-head among the super-nodes can connect directly at least one of the other cluster-heads for guarantee of strong connection. The greedy heuristic clustering tries to partition a large scale ad-hoc network into clusters. Simultaneously, it adjusts all nodes power level for the purpose of power conservation. The selected cluster-heads comprise the super-nodes topology which MPGC can execute on. Importantly, the cluster structure could be disturbed due to the mobility of nodes. K. The Probability for Rening Energy-Efciency of Multicast Tree Protocol (P-REMiT) The Probability for Rening Energy-efciency of Multicast Tree (P-REMiT) [44] is a tree-based multicast routing protocol for building an energy efcient multicast routing. It uses the probability method to balance the total energy consumption and system lifetime of multicast tree. P-REMiT assumes that nodes with omni-directional antennas are stationary, and each node knows the distance between itself and its neighboring nodes. P-REMiT includes three major steps as follows: (1) Building an initial multicast tree; all nodes run the distributed algorithm to build a multicast tree. (2) Rening the multicast tree; the phase is organized in rounds. Each round is led by the multicast source nodes. Source nodes terminate P-REMiT when there is no change in the last round. (3) Eliminating all non-members redundant transmissions by pruning the multicast tree. L. The Power-Aware Multicast Routing Protocol (PMRP) The power-aware multicast routing protocol (PMRP) [45] is a tree-based minimum energy multicast routing protocol

JUNHAI et al.: A SURVEY OF MULTICAST ROUTING PROTOCOLS FOR MOBILE AD-HOC NETWORKS

87

with mobility prediction. In order to select a subset of paths that provide increased stability and reliability of routes in routing discovery, each node receives the RREQ packet and uses the power-aware metric to get in advance the power consumption of transmitted data packets. If the node has enough remaining power to transmit data packets, it uses the GPS to get the location information (i.e. position, velocity and direction) of the MNs and utilizes this information to calculate the Link Expiration Time (LET) between two connected MNs. During routing discovery, each destination node selects the routing path with the smallest LET and uses this smallest link expiration time as the Route Expiration Time (RET). The destinations nodes collect several feasible routes and then select the path with the longest RET as the primary routing path. Then the source node uses these routes between the source node and each destination node to create a multicast tree. In the multicast tree, the source node will be the root node and the destination nodes will be the leaf nodes. M. The CodeCast Multicast Routing Protocol (CCMRP) The CodeCast Multicast Routing Protocol (CCMRP) [46] is a network-coding, controlled-loss, bounded-delay-based multicast routing protocol, which is random network coding and transparently implements both localized loss recovery and path diversity with very low overhead. It is suited for multimedia applications and multimedia data is delivered by a multicast source. The main ingredient of CCMRP is random network coding, which is used to implement both localized loss recovery and path diversity transparently. CCMRP can achieve a near-perfect packet delivery ratio while maintaining lower overhead than conventional multicast. CCMRP assumes an application generates a steam of equal size frames that can uniquely distinguished by source address and port-number pair or a globally unique identication number assigned to each stream. a priori and possibly resource allocations. But the use of CCMRP is limited due to those assumptions. N. The On Network Coding and Routing in Dynamic Wireless Multicast network (ONCRM) The On Network Coding and Routing in dynamic wireless Multicast network (ONCRM) [47] is a power-aware multicast routing protocol using network coding for MANETs. ONCRM is optimal for intra-session network coding and routing based on dynamic back pressure protocols. It uses queue state information to make network coding and transmit scenario decisions, without requiring any knowledge of the input or channel statistics and each node maintains three queues for every subset of sinks: two individual queues containing data that is to be transmitted to each of the sinks and a common queue containing data that is to be transmitted to both sinks. The following ve steps (i.e. balancing, scheduling, network coding, power control and routing) are carried out in each time slot in ONCRM. O. The Distributed QoS Multicast Routing Protocol (DQMRP) The Distributed QoS Multicast Routing Protocol (DQMRP) [48] is a shared-tree QoS-based multicast routing protocol.

The multicast tree is formed incrementally by source node. Source node sends an explored frame, which recodes every intermediate node it passes, including source node, to every neighboring node with feasible path from itself. And the intermediate node transfers this frame received within limited time to all its neighbor nodes expect source node. If it receives over the constrained time, it discards any frame. The destination node chooses the path with minimum cost from more than on feasible paths. At the same time, destination node reversesends a resource reservation information and acknowledge reply to source nodes and adds it into the multicast tree. Then destination node keeps all other feasible paths recorded by other explorer frame as backup paths. Acknowledge reply nds the path to the source nodes through the previous nodes information kept among the intermediate nodes, meanwhile, the intermediate nodes update their previously nodes information through the source of the acknowledge reply information. When resource reservation information or acknowledge reply information nally reaches source node, the destination node is added into the multicast tree successfully. P. The Logical Hypercube-Based Virtual Dynamic Backbone protocol (HVDB) The logical Hypercube-based Virtual Dynamic Backbone (HVDB) [49] is a proactive, QoS-aware and hybrid multicast routing protocol for large scale MANETs. It includes proactive logical route maintenance, summary-based membership update and logical location-based multicast routing. Due to the regularity and symmetry properties of hypercube, no leader is needed in a logical hypercube, and every node plays almost the same role except for the slightly different roles of border cluster heads and inner cluster heads. Thus, no single node is more loaded than any other nodes, and no problem of bottlenecks exists, which is likely to occur in tree-based architectures .HVDB supports QoS-aware multicast and is derived from n-dimensional hypercube in large-scale MANETs, which have many desirable properties, such as high fault tolerance, small diameter, regularity, and symmetry. It uses the location information of MNs and meets the new QoS requirements: high availability and good load balancing. VI. A S UMMARY OF M ULTICAST ROUTING P ROTOCOLS We summarize the classication of the multicast routing protocols discussed in this paper in a table (see Table 1). The table lists the multicast routing philosophies (e.g., proactive, reactive, at, hierarchical, location-awareness and powersensitiveness etc), the underlying multicast routing metrics and the primary routing selection principles of those protocols. VII. C ONCLUSION AND F UTURE W ORK A. Conclusion In summary, multicasting can efciently support a wide variety of applications that are characterized by a close degree of collaboration, typical for many MANETs. And the design of the multicast routing protocols are driven by specic goals and requirements based on respective assumptions about the network properties or application areas. In this paper, we present

88

IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 11, NO. 1, FIRST QUARTER 2009

TABLE I S UMMARY OF M ULTICAST ROUTING P ROTOCOLS

Protocol/ Algorithm Acronym/ Name AMRIS [10] MAODV[11] AMRoute[12] LAM [13] ODMRP [14] CAMP [15] LGT [16] DDM[17] HQMRP[18] SOM[19] LGF[22] SPBM[23] STAMP[24] ACMP[25] CQMP[26] E-ODMRP[27] BODS[28] EHMRP[29] MWIA[32] PCHMR[33] RAMP[37] RORP[38] ROMANT[39] OPHMR[41] MAMR[42] MPGC[43] P-REMiT[44] PMRP [45] CCMRP[46] ONCRM[47]

HY

Primary Multicast Routing Metric

Multicast routing selection philosophy AI Y Y Y Y Y Y Y Y N N Y N N Y Y Y Y Y N N N N N N N N N N N N N N AD N N N N N N N N Y Y N Y Y N N N N N Y Y Y Y Y Y Y Y Y Y Y Y Y Y

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

N N N N N N N N N N N N N N N N N N Y Y N N N N N Y Y Y N N

Y Y Y Y Y Y Y Y N N Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y

N N N N N N N N Y Y N N N N N N N N N N N N N N N N N N N N

N N N N N N N Y N N N N N N N N N Y N N N N N Y N N N N N N

N N N N N N N N N N Y Y N N N N N N N Y N N N N N N N N N N

N N N N Y Y N N N N Y N N N Y Y Y N N N Y N N N N N Y Y N N

Y N Y N Y Y N Y Y N N N N N N N N Y Y Y Y Y N N N Y Y Y Y Y

N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N

N Y N Y N N Y N N Y Y Y Y Y Y Y Y N Y N N N Y Y Y N N N N N

Y Y Y N N N N N Y Y N N Y Y Y N Y Y Y N N Y Y N N Y Y Y Y Y

Y Y Y Y N N Y

31 DQMRP[48] N Y N N N N Y Y N Y 32 HVDB[49] N N Y Y N N Y Y N Y AI-Application Independence AD-Application Dependence E-Energy-aware H-Hierarchical HY-Hybrid L-Location-aware M-Mesh P-Proactive Q-QoS-aware R-Reactive S-Stability T-Tree

Minimum hop Minimum hop Minimum hop Link affinity Minimum hop Link affinity Maximum Forward Progress N Minimum hop N Localized reactions to topology changes N Link affinity N Maximum Forward Progress Y Maximum forward progress T Max. forward progress/ Minimum hop Y Max. forward progress/ Minimum hop Y Minimum hop Y Minimum hop Y Minimum hop Y Max. forward progress/ Minimum hop N Minimum overall end-to-end transmission power Y Minimum transmission power N Minimum hop Y Max. forward progress/ Minimum hop Y Max. forward progress/ Minimum hop N Minimum hop Y Min. hop path on stronger stable channels Y Minimum overall end to-end transmission delay Y Minimum transmission power N Minimum end-to-end transmission power N Min. hop path on stronger stable channels N Min. hop path on stronger stable channels Y Maximum forward progress Y Maximum forward progress F-Flat

a comprehensive survey of the multicast routing protocols for MANETs. The purpose of this paper is to survey the multicast routing protocols and study their primary routing selection principles.We discuss the characteristics, routing metrics and routing philosophies of each of these protocols selected from the class of similar approaches ,which can reect the stateof-the-art research work on multicast routing protocols. The classications of the primary routing selection principles can simplify the task of a network designer in deciding the multicast routing strategies to be adopted at a given condition. Then, we believe our survey will be very useful to the research

community and also serve as a great introductory material for someone embarking on MANETs. B. Future Work As mentioned earlier, research in the area of multicast routing protocols over MANETs is far from comprehensive. Much of the effort so far has been on devising multicast routing protocols to support effective and efcient communication between nodes that are part of a multicast group [50]. However, there are still many topics that deserve further investigation as follows:

JUNHAI et al.: A SURVEY OF MULTICAST ROUTING PROTOCOLS FOR MOBILE AD-HOC NETWORKS

89

TABLE II E XPLANATION OF M ULTICAST ROUTING P ROTOCOL A CRONYMS

# 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Protocol/ Algorithm Acronym / Name AMRIS[10] MAODV[11] AMRoute[12] LAM[13] ODMRP[14] CAMP[15] LGT[16] DDM[17] HQMRP[18] SOM[19] LGF[22] SPBM[23] STAMP[24] ACMP[25] CQMP[26] E-ODMRP[27] BODS[28] EHMRP[29] MWIA[32] PCHMR[33] RAMP[37] RORP[38] ROMANT[39] OPHMR[41] MAMR[42] MPGC[43] P - R E M iT [ 4 4 ] PMRP[45] CCMRP[46] ONCRM[47] DQMRP[48] HVDB[49]

Protocol/ Algorithm Name Ad-hoc Multicast Routing protocol utilizing Increasing id numbers protocol Multicast Ad-hoc On-Demand Vector protocol Ad-hoc Multicast Routing protocol Lightweight Adaptive Multicast algorithm On-Demand Multicast Routing Protocol Core Assisted Mesh Protocol Location Guided Tree algorithm Differential Destination Multicast alg o rith m Hierarchical QoS Multicast Routing Protocol Self-Organizing Map algorithm Location-based Geocasting and Forwarding algorithm Scalable Position-Based Multicast algorithm Shared Tree Ad-hoc Multicast Protocol Adaptive Core-based Multicast Routing Protocol Mesh-based Multicast Routing Protocol with Consolidated Query Packets Enhanced On-Demand Multicast Routing Protocol Bandwidth Optimized and Delay Sensitive algorithm Efficient Hybrid Multicast Routing Protocol Minimum Weight Incremental Arborescence algorithm Power-Controlled Hybrid Multicast Routing protocol Reliable Adaptive Multicast Protocol Reliable On-demand Routing Protocol RObust Multicasting in Ad-hoc Network using Tree Optimized Polymorphic Hybrid Multicast Routing protocol Mobile Agents aided Multicast Routing protocol Multicast Power Greedy Clustering protocol Probability for Refining Energy-efficiency of Multicast Tree Dis tr ib uted Mini mu m Ener g y Multic a st CodeCast Multicast Routing Protocol On Network Coding and Routing in dynamic wireless Multicast network Distributed QoS Multicast Routing Protocol logical Hypercube-based Virtual Dynamic Backbone

1) Scalability; this issue is not only related to multicast routing protocols but also with the mobile ad-hoc network itself. Can we design a multicast routing protocol that is scalable with respect to number of members in the group, their mobility, and other constraints posed by the ad-hoc environment itself? 2) Address conguration; in ad-hoc environment, a different addressing approach may be required. Special care must be taken so that other groups do not reuse a multicast address used by a group at the same time. Node movement and network partitioning makes this task of synchronizing multicast addresses in ad-hoc really difcult. 3) Multicast service support; the multicast protocol denes conditions for joining/leaving groups, multicast participants should be able to join or leave groups at will. On the other hand, whether service providers can be convinced to support multicast is still an open issue. 4) Security; how can the network secure itself from malicious or compromised nodes? Due to the broadcast nature of the wireless medium security provisioning becomes more difcult. Further research is needed to investigate how to stop an intruder from joining an

5)

6)

7)

8)

ongoing multicast session or stop a node from receiving packets from other sessions. Trafc control; both source and core-based approaches concentrate trafc on a single node. In stateless multicast group membership is controlled by the source, which leads to the vulnerability of multicast protocols. Still need to be investigated is how to efciently distribute trafc from a central node to other member nodes. QoS; QoS denes a guarantee given by the network to satisfy a set of predetermined service performance constraints for the user in terms of end-to-end delay, jitter, and available bandwidth. Therefore, multicast routing protocols must be feasible for all kinds of constrained multicast applications to run well in MANETs. However, it is a signicant technical challenge to dene a comprehensive framework for QoS support, due to dynamic topology, distributed management and multihop connections. Power control; for power-constrained wireless networks, a crucial issue in routing and multicasting is to conserve as much power as possible while still achieving good throughput performance. Multiple Source; Most of the existing multicast routing

90

IEEE COMMUNICATIONS SURVEYS & TUTORIALS, VOL. 11, NO. 1, FIRST QUARTER 2009

protocols in MANETs are designed for single source multicasting. However, a multicast group may contain multiple sources due to different kinds of services or applications simultaneously provided by the networks. Each single source multicast routing protocol induces a lot of overhead and thus wastes tremendous network resources in multi-source multicast environment. ACKNOWLEDGEMENT The authors wish to thank the reviewers and the editors for their valuable suggestions and comments that help improve the paper. R EFERENCES
[1] H. Deng, W. Li, and D. P. Agrawal, Routing security in wireless adhoc networks, IEEE Commun. Mag., vol. 40, no. 10, pp. 7075, Oct. 2002. [2] M. Younis and S. Z. Ozer, Wireless ad-hoc networks: Technologies and challenges, Wireless Commun. Mobile Computing, vol. 6, no. 7, pp. 889892, Nov. 2006. [3] S. Guo and O. Yang, Energy-aware multicasting in wireless ad-hoc networks: A survey and discussion, Computer Commun., vol. 30, no. 9, pp. 21292148, June 2007. [4] X. Chen and J. Wu, Multicasting techniques in mobile ad-hoc networks, The Handbook of Ad-hoc Wireless Networks, pp. 2540, 2003. [5] L. Junhai and Y. Danxia, et al., Research on routing security in MANET, Application Research of Computers, vol. 25, no. 1, pp. 243 245, Jan. 2008. [6] A. Shaikh, J. Rexford, and K. G. Shin, Evaluating the impact of stale link state on quality-of-service routing, IEEE/ACM Trans. Networking, vol. 9, no. 2, pp. 162176, Apr. 2001. [7] J.-S. Chou, C.-H. Lin, and C.-H. Chiu, An identity-based scheme for ad-hoc network secure routing protocol from pairing, WSEAS Trans. Computers, vol. 5, no. 6, pp. 12141221, June 2006. [8] C. Liu, Y. Shu, and Y. Zhou, et al., A comparison of DSR, MSR and BSR in wireless ad-hoc networks, SPIE, vol. 6011, pp. 601610, 2005. [9] M. K. Marina and S. R. Das, Ad-hoc on-demand multi-path distance vector routing, Wireless Commun. Mobile Computing, vol. 6, no. 7, pp. 969988, Nov. 2006. [10] C. W. Wu and Y. C. Tay, AMRIS: A multicast protocol for ad-hoc wireless networks, in Proc. IEEE MILCOM, 1999, vol. 1, pp. 2529. [11] E. M. Royer and C. E. Perkins, Multicast operation of the ad-hoc ondemand distance vector routing protocol, in Proc. ACM MOBICOM, Aug. 1999, pp. 207218. [12] J. Xie, R. Rajesh, and A. McAuley, et al., AMRoute: Ad-hoc multicast routing protocol, Mobile Networks and Applications, Multipoint Communication in Wireless Mobile Networks, vol. 7, no. 6, pp. 429439, Dec. 2002. [13] J. J. Garcia-Luna-Aceves and E. L. Madruga, Core-assisted mesh protocol, IEEE J. Select. Areas Commun., vol. 17, no. 8, pp. 1380 1394, 1999. [14] M. Gerla, S. J. Lee, and W. Su, On-demand multicast routing protocol (ODMRP) for ad-hoc networks, Internet draft, draft-ietf-manet-odmrp02.txt, 2000. [15] L. Ji and M. S. Corson, A lightweight adaptive multicast algorithm, in Proc. IEEE GLOBECOM, 1998, pp. 10361042. [16] K. Chen and K. Nahrstedt, Effective location-guided tree construction algorithms for small group multicast in MANET, in Proc. IEEE INFOCOM, 2002, vol. 3, pp. 11801189. [17] L. Ji and M. S. Corson, Differential destination multicast-A MANET multicast routing protocol for small groups, in Proc. IEEE INFOCOM, 2001, vol. 2, pp. 11921201. [18] L. Li and C. Li, A hierarchical QoS multicast routing protocol for mobile ad-hoc networks, Chinese J. Electronics, vol. 15, no. 4, pp. 573577, Oct. 2006. [19] M. Kumar, V. Senthil, C. Venkatesh, and A. M. Natarajan, Performance comparison of multicast protocol for physically hierarchical ad-hoc networks using neural concepts, ICSP, vol. 2, pp. 15811584, 2004. [20] H. Ammari, El-Rewini, and Hesham, A multicast protocol for mobile ad-hoc networks using location information, ASWN, pp. 174183, 2004. [21] R. Friedman and G. Kliot, Location services in wireless ad-hoc and hybrid networks: A survey, Department of Computer Science, Technion, Haifa, Israel, no. 4, pp. 115, Apr. 2006.

[22] L. A. Latiff, A. Ali, and C.-C. Ooi, Location-based geocasting and forwarding (LGF) routing protocol in mobile ad-hoc network, AICT/SAPIR/ELETE, pp. 536541, 2005. [23] M. Transier, H. Fussler, and J. Widmer, et al., A hierarchical approach to position-based multicast for mobile ad-hoc networks, Wireless Networks, vol. 13, no. 4, pp. 447460, Aug. 2007. [24] L. Canourgues, J. Lephay, and Soyer, et al., STAMP: Shared-tree adhoc multicast protocol, in Proc. IEEE MILCOM, Oct. 2006, pp. 17. [25] B. Kaliaperumal, A. Ebenezer, and Jeyakumar, Adaptive core-based scalable multicasting networks, in Proc. IEEE INDICON, Dec. 2005, pp. 198202. [26] H. Dhillon and H. Q. Ngo, CQMP: A mesh-based multicast routing protocol with consolidated query packets, in Proc. IEEE WCNC, 2005, vol. 4, pp. 21682174. [27] Y. O. Soon, J.-S. Park, and M. Gerla, E-ODMRP: Enhanced ODMRP with motion adaptive refresh, in Proc. ISWCS, 2005, pp. 130134. [28] E. R. Inn Inn and W. K. Seah, Distributed steiner-like multicast path setup for mesh-based multicast routing in ad-hoc networks, IEEE TIME, pp. 192197. [29] J. Biswas, M. Barai, and S. K. Nandy, Efcient hybrid multicast routing protocol for ad-hoc wireless networks, IEEE LCN, pp. 180187, Nov. 2004. [30] L. Layuan and L. Chunlin, A QoS multicast routing protocol for clustering mobile ad-hoc networks, Computer Communications, vol. 30, no. 7, pp. 113. [31] L. Jiageng, D. Cordes, and J. Zhang, Power-aware routing protocols in ad-hoc wireless networks, IEEE Wireless Commun., vol. 12, no. 6, pp. 6981, Dec. 2005. [32] W. Cheng, C. Wen, and K. Feng, Power-controlled hybrid multicast routing protocol for mobile ad-hoc networks, in Proc. IEEE VTC, 2006, vol. 3, pp. 10871091. [33] M. X. Cheng and J. Sun, et al., Energy-efcient broadcast and multicast routing in multi-hop ad-hoc wireless networks Wireless Communications and Mobile Computing, vol. 6, no. 2, pp. 213223, Mar. 2006. [34] A. Davidian, Network coding: An overview, Institute for Communications Engineering (LNT), vol. 13, no. 7 pp. 119, Jan. 2005. [35] D. S. Lun, Ahmed, and Ebad, et al., Network coding for wireless applications, in Proc. SPIE, 2005, pp. 258260. [36] Y. C. Wu, A. Philip and S.-Y. Kung, Minimum-energy multicast in mobile ad-hoc networks using network coding, IEEE Trans. Commun., vol. 53, no. 11, pp. 19061918, Nov. 2005. [37] B. Sun and L. Li, Reliable adaptive multicast protocol in wireless adhoc networks, J. Systems Engineering Electronics, vol. 17, no. 1, pp. 187192, Mar. 2006. [38] N.-C. Wang and S.-W. Chang, A reliable on-demand routing protocol for mobile ad-hoc networks with mobility prediction, Computer Commun., vol. 29, no. 1, pp. 123135, Dec. 2005. [39] R. Vaishampayan and J. J. Garcia-Luna-Aceves, Robust tree-based multicasting in ad-hoc networks performance, in Proc. IEEE IPCCC, 2004, vol. 23, pp. 647652. [40] L. K. Law, S. V. Krishnamurthy, and M. Faloutsos, A novel adaptive protocol for lightweight efcient multicasting in ad-hoc networks, Computer Networks, vol. 1, no. 3, pp. 823834, Feb. 2007. [41] A. B. Mnaouer, L. Chen, and C. H. Foh, et al., OPHMR: An optimized polymorphic hybrid multicast routing protocol for MANET, IEEE Trans. Mobile Computing, vol. 6, no. 5, pp. 503514, May 2007. [42] H. M. P. Shekhar, M. A. Arun Kumar, and K. S. Ramanatha, Mobile agents aided multicast routing in mobile ad-hoc networks, in Proc. IEEE ICACT, 2005, vol. 2, pp. 765770. [43] J. J. Y. Leu, M.-H. Tsai, and C. Tzu-Chiang, et al., Adaptive poweraware clustering and multicasting protocol for mobile ad-hoc networks, UIC, vol. 2006, pp. 331340. [44] Y. Luo, J. Wang, and J. Chen, et al., Algorithm based on mobility prediction and probability for energy-efcient multicasting in ad-hoc networks Computer Research and Development, vol. 43, no. 2, pp. 231237, Feb. 2006. [45] G. Song, V. Leung, and O. Yang, A distributed minimum energy multicast algorithm in MANETs, in Proc. WoWMoM, vol. 2006, pp. 134140. [46] J.-S. Park, M. Gerla, and D. S. Lun, et al., CodeCast: A networkcoding-based ad-hoc multicast protocol, IEEE Trans. Wireless Commun., vol. 13, no. 5, pp. 7681, Oct. 2006. [47] T. Ho, B. Leong, and M. Medard, et al., On the utility of network coding in dynamic environments, in Proc. International Workshop Wireless Ad-Hoc Networks, 2005, pp. 196200. [48] B. Sun and L. Li, Distributed QoS multicast routing protocol in adhoc networks, J. Systems Engineering Electronics, vol. 17, no. 3, pp. 692698, Sep. 2006.

JUNHAI et al.: A SURVEY OF MULTICAST ROUTING PROTOCOLS FOR MOBILE AD-HOC NETWORKS

91

[49] G. Wang, C. Guojun, and L. Zhang, et al., A novel QoS multicast model in mobile ad-hoc networks, in Proc. IEEE IPDPS, 2005, p. 206. [50] J. Luo and D. Ye, et al., Research on topology discovery for IPv6 networks, in Proc. IEEE SNPD, 2007, vol. 3, pp. 804809.

Luo Junhai received the BTech degree in computer science and appliance from University of Electronic Science and Technology of China in 2003, the MTech degree in computer appliance technology from Chengdu University of Technology in Chengdu P. R. China in 2006. He is a visiting scholar at McGill University at Canada in 2008. He is presently pursuing a PhD degree in information security at University of Electronic Science and Technology of China. His research interests and papers are mostly in the areas of information security, DSP and wireless communication including ad hoc network, mesh network, TCP over wireless and sensor network. He is a member of the IEEE and the CCF.

Xue Liu received the BS degree in applied mathematics and the MEng degree in control theory and applications from Tsinghua University in 1996 and 1999, respectively. He received his PhD degree from University of Illinois at Urbana Champaign. He is a professor at McGill University in Canada His research interests include real time and embedded computing, performance and power management of server systems, sensor networks, fault tolerance, and control. He has authored/coauthored more than 20 refereed publications in leading conferences and journals in these elds. He received the Ray Ozzie Fellowship, the Saburo Muroga Fellowship, the C.W. Gear Outstanding Graduate Award, and the Mavis Memorial Fund Scholarship Award from the University of Illinois at Urbana Champaign. He is a member of the IEEE.

Fan Mingyu is a professor. Her research direction is information security, multimedia computing, Web services and grid computing, spatial information processing, medical informatics, logistics and e-commerce, information hiding and data warehouse.

Ye Danxia is working toward her master degree with school of software of University of Electronic Science and Technology of China. Her research directions are information security, computer network and software engineering.