This action might not be possible to undo. Are you sure you want to continue?
A survey on quality-of-service support for mobile ad hoc networks
Dmitri D. Perkins*,† and Herman D. Hughes Department of Computer Science and Engineering Michigan State University East Lansing MI 48824-1027 U.S.A. Summary The general ﬁeld of mobile ad hoc networking is still in its infancy. Particularly, the challenge of providing Quality-of-Service (QoS) support for ad hoc networks is an open problem and remains relatively uncharted territory. Providing a complete QoS solution for the ad hoc networking environment requires the interaction and cooperation of several components. These components include: (1) a QoS routing protocol, (2) a resource reservation scheme and (3) a QoS capable medium access control (MAC) layer. In this paper, we present a survey of the current research that has addressed each of these components in the context of ad hoc networks. This work is intended to provide a broad and comprehensive view of the various components and protocols required to provide QoS support in computer networks, focusing primarily on ad hoc networks. First, we introduce the unique characteristics of mobile ad hoc networks, which distinguishing this new network architecture from traditional infrastructured wired and wireless networks (i.e. cellular-based networks). We also discuss the impact of these characteristics on QoS provisioning. Next, we describe the ﬁrst QoS model proposed for mobile ad hoc networks and its relationship to QoS models proposed for the Internet. We then present a review of the proposed algorithms for each QoS component (e.g. QoS routing, resource reservation and the MAC layer). Copyright 2002 John Wiley & Sons, Ltd.
KEY WORDS Quality-of-Service (QoS) wireless networks mobile ad hoc networks QoS routing, resource reservation QoS-base medium access control (MAC)
Correspondence to: Dmitri D. Perkins, Department of Computer Science and Engineering, Michigan State University, East Lansing, MI 48824-1027, U.S.A. † E-mail: email@example.com Copyright 2002 John Wiley & Sons, Ltd.
the network topology may change randomly and rapidly and may consist of both bidirectional and unidirectional links. and must cooperate in order to dynamically establish communications using limited network management and administration . which differ from traditional wireless systems. Mobility is managed by allocating a limited set of communications frequency channels to each BS. They are composed of wireless mobile nodes (e.504 D. Such challenges include. All communications. and dynamically assigning a mobile device to a local channel as it moves from the coverage area of one BS to another. Introduction The emergence of real-time applications and the widespread use of wireless and mobile devices have generated the need to provide quality-of-service (QoS) support in wireless and mobile networking environments. This is a difﬁcult task for best-effort routing without even considering QoS requirements. bandwidth. are free to move rapidly and arbitrarily in time and space. supporting QoS in mobile ad hoc networks. Mobile ad hoc Networks: Properties and Design Challenges ad hoc networks are self-organizing. including energy. is often much less than a radio’s maximum transmission rate. HUGHES 1. Thus. particularly where mobile access to a wired network is either ineffective Copyright 2002 John Wiley & Sons. 2002. Resources. the realized throughput of wireless communications—after accounting for the effects of multiple access. including user terminals. congestion control and. Furthermore. a mobile ad hoc network is a network of mobile routers. Wirel. Ltd. ad hoc networks will emerge as an effective complement to wired or wireless LANs. 2. or stationary. 5]. Mob. The limited power of the mobile nodes and the lack of a ﬁxed infrastructure in ad hoc networks restrict the transmission range. processing capacity and memory. rapidly deployable and require no ﬁxed infrastructure [1–3]. consisting of bandwidth-constrained links. QoS support. making it difﬁcult to determine the aggregate bandwidth between two endpoints. Wireless links will continue to have signiﬁcantly lower capacity than their hardwired counterparts.g. Communication from that point is routed across a ﬁxed network to its destination. In addition. but are not limited to. the primary focus of this paper. They may operate autonomously or connected to the Internet. fading. however. and supporting a wide range of networking applications. ad hoc networks have the potential to serve as a ubiquitous wireless infrastructure capable of interconnecting thousands of devices . Essentially. PERKINS AND H. a router consisting of multiple hosts and equipped with wireless communication capability) that can be deployed anywhere.g. applications and services equivalent to those available in these environments must be made available to ad hoc network users. which are the basis for these challenges  1. is even more difﬁcult. and may vary widely in terms of their capabilities and uses [4. There is no centralized authority for network control. Comput. A set of ﬁve (5) properties are listed below. D. noise and interference conditions and so on. The remainder of this section provides an overview of ad hoc networks including its distinguishing features and associated challenges. effective multihop routing. and even to wide-area mobile networking services. BS). 2:503– 513 . or impossible. 3. user data and control information. Network devices. It is hoped that in the future. routers and other potential service platforms. The successful implementation of mobile ad hoc networking technology presents a unique set of challenges. Current wireless networks support mobile/wireless access for mobile communications devices by providing a wireless interface between the mobile devices and a ﬁxed network of limited range base-stations (BS). Because of their inherent ﬂexibility. mobility and data management.1. This paper provides a survey of current research concerned with the problem of providing QoS in a wireless network architecture called mobile ad hoc networks. which do not depend on a BS for communications. In order to achieve this status. these effects will also result in time-varying channel capacity. The section concludes with a brief overview of QoS concepts and deﬁnitions. Nodes in an ad hoc network may be highly mobile. are strictly limited and must be preserved. the air-interface consists of a single data-link terminating on a BS. routing or administration (e. D. While providing QoS in an infrastructured environment is difﬁcult. such as Personal Communication Systems (PCS). are carried over the wireless medium. Environments in which ad hoc networks are initially expected to play an important role include instant infrastructure scenarios. MAC. resulting in the need for effective multihop routing in ad hoc networks. On the basis of this infrastructured model for wireless communications. Commun. 1. 4. that are relatively abundant in wired environments.
In this paper. these enhancements may not be suitable for ad hoc networks. Comput.g.2. For example. Mobile nodes that are end points for user communications and applications must act cooperatively to handle network functions. estimated delay etc. bandwidth. while providing QoS support in addition to ﬂexibility and mobility is a tremendously challenging task for the Internet as well as cellular networks. packet loss rate. bandwidth etc. For example. Ltd. 2002. The associated QoS could. Technically. tries to utilize resources efﬁciently and to make the network QoSaware. Typical QoS metrics include available bandwidth. network designers could simply increase the capacity of a Copyright 2002 John Wiley & Sons. However. Protocols such as routing. congestion link or network from 10 to 100 Mb.) and topology information.g. This includes (1) QoS-based routing protocols [8–11]. buffers etc. resource reservations and so on.g. While adaptive algorithms are required for the successful implementation of ad hoc networks. videoconferencing and Internet telephony have. a QoS model does not deﬁne speciﬁc protocols or implementations. The cost associated with this must be considered since control packets will also compete with data packets for network resources (e. packets are dropped regardless of their importance. followed by a summary in Section 6. which addresses the problem of supporting QoS in ad hoc networks. are sensitive to packet loss and delay and may have minimum bandwidth requirements Consequently. This could include additional service classes. Instead. Analogous to today’s Internet. available bandwidth. The second approach. we present a review of currently proposed QoS models for ad hoc networks. because of node movement. QoS is a set of service requirements to be met by the network while transporting a ﬂow.g. QoS MAC layer protocols are presented in Section 5. 2. control overhead is also a key consideration. be ‘best effort’. digitized audio and video).). However. there are two ways in which QoS can be achieved: (1) over-provisioning and (2) trafﬁc engineering. packet loss rate. limited resources and node mobility make maintaining accurate routing information very difﬁcult if not impossible in an ad hoc networking environment. These properties and their associated challenges will potentially impair any attempt to provide QoS guarantees in ad hoc networks. The timevarying capacity of wireless links. As stated in property four. A ﬂow is a packet stream from a source to a destination (unicast or multicast) with an associated (QoS). in fact. it deﬁnes the methodology and architecture by which certain types of services (e. Recently proposed QoS routing protocols are presented in Section 3 including the deﬁnitions of the QoS routing classes. channel access. such as video-on-demand (VoD). If a packet is lost. QoS Service Models for Mobile ad hoc Networks Generally. estimated delay. without specialized routers. A fundamental requirement of any QoS mechanism is a measurable performance metric. (2) resource reservation schemes [12–15] and (3) a MAC protocol [16. hop count and path reliability. The remainder of this paper is organized as follows. these components work together to achieve speciﬁc goals that are speciﬁed by a QoS service model [18–20]. Mob. admission control.). the sender can simply retransmit the lost packet. real-time applications. 2:503– 513 . For example. do not provide any guarantees regarding packet loss or delay. Thus. we focus on the trafﬁc-engineering approach. connections in ad hoc networks may need to be reestablished.A SURVEY ON QOS SUPPORT FOR MOBILE NETWORKS 505 5. 1. In the next section. 17]. In a best-effort service model. packet jitter. available bandwidth. jitter etc. The purpose of this paper is to provide a broad survey on the current research. per-ﬂow or class-based) can be provided in the network. current QoS routing algorithms require accurate link state (e. This method is efﬁcient for applications that do not require bounds on packet delay or other QoS metrics. trafﬁc engineering.g. mostly notably routing and MAC. the best-effort service may not be suitable for these applications.e. Quality-of-Service Provisioning: Deﬁnition and Overview According to RFC2386 . Overprovisioning utilizes the best-effort approach and simply increases the available resources (e. Section 4 discusses the current work pertaining to signaling and resource reservation in ad hoc networks. in which the mobile node is only a single hop from a wired ﬁxed infrastructure. ad hoc networks are being designed to provide best-effort service (i. supporting QoS in ad hoc networks is an even more difﬁcult challenge. these resources are very limited in ad hoc networks and must be preserved for data trafﬁc. Commun. Wirel. Generally.). Research and development efforts are under way to enhance the Internet with QoS components that will allow the transport of real-time data (e.
In FQMM. FQMM consist of three key features: dynamic roles of nodes. The actual state information could include bandwidth requirements. IntServ aims to emulate a connection-oriented. Since state information is maintained only for a small portion of trafﬁc. The QoS routing protocol tries to ﬁnd a path that has a good chance of meeting the QoS requirements . In contrast to an absolute trafﬁc proﬁle. A QoS model designed for ad hoc networks must consider the unique features and challenges associated with mobile ad hoc networks—in particular. The provisioning approach in FQMM consists of a hybrid per-ﬂow (IntServ) and per-class (DiffServ) scheme Copyright 2002 John Wiley & Sons. a single host may have multiple roles. 2002. Adaptive conditioning. As such.g. Two QoS models. a path that has sufﬁcient available resources capable of satisfying the QoS requirements). Hybrid provisioning. packet delay and loss bounds or delay variation. which polices the trafﬁc according to the trafﬁc proﬁle and is responsible for marking the trafﬁc streams and discarding packets. The QoS Wirel.e. Provisioning is used to determine and allocate needed resources at various points in the networks—these points are mobile host in Mobile ad hoc Networkings (MANETs). Commun. which have been proposed for the Internet. QoS routing is a routing mechanism under which paths for ﬂows are determined on the basis of some knowledge of resource availability in the network as well as the QoS requirements of the ﬂows or connections . The DiffServ architecture is intended to provide scalable service differentiation in the Internet without the need for maintaining per-ﬂow state information and signaling at every router. node mobility (dynamic topology) and time-varying link capacity. Designers of QoS routing algorithms for ad hoc networks must consider several design issues: (1) metric selection (e. D. 2:503– 513 Dynamic roles of nodes. Before any connections can be made or any resources reserved. the scalability problem of IntServ is expected to improve. ﬁnd a feasible path between a sourcedestination pair (i. in which trafﬁc of the highest-priority is given perﬂow treatment.e. This approach requires maintaining speciﬁc state information for every ﬂow in every router. (2) an interior node is a node that forwards data for other nodes and (3) an egress node is a destination node. PERKINS AND H. the trafﬁc proﬁle proposed in FQMM is deﬁned as the relative percentage of the effective link capacity. are Integrated Services (IntServ) and Differentiated Services (DiffServ). FQMM deﬁnes three types of nodes: (1) an ingress node is a mobile node that sends data. D. Comput. an application does not explicitly signal the network (i. This model is called Flexible QoS Model for Mobile ad hoc Networks (FQMM) and is a hybrid of the IntServ and DiffServ service models. Since nodes are free to move resulting in topology changes. the routers) before transmitting data. Diffserv proposes a service model and algorithms to support QoS for aggregated trafﬁc classes. The remainder of this section describes the ﬁrst QoS service model designed speciﬁcally for ad hoc networks. IntServ and DiffServ were proposed for static networks and thus cannot be applied directly to the mobile ad hoc environment. bandwidth. The trafﬁc conditioner. The adaptive trafﬁc conditioner includes several components: a trafﬁc proﬁle. (2) QoS state propagation and maintenance and (3) scalability. marker and dropper. Ltd. while other trafﬁc is given per-class provisioning. is placed at the ingress node at which the trafﬁc originates. preserving the per-ﬂow granularity for a small portion of trafﬁc in MANETs. thus. providing smooth performance degradation to lower-priority trafﬁc. delay etc. .506 D. meter. bandwidth allocation is used as the relative service differentiation parameter. the network tries to deliver a particular kind of service based on the QoS speciﬁed by each packet. hybrid provisioning and adaptive conditioning. HUGHES resource reservation/signaling and MAC must cooperate to achieve the goals outlined by the QoS model. QoS Routing in Mobile ad hoc NETworks QoS routing is an essential part of the QoS architecture.) and path computation. The objectives of QoS routing are threefold: (1) if one exists. a feasible path between a sourcedestination pair must be established. FQMM assumes that the larger proportion of trafﬁc does not belong to the highest-priority class. Mob. Under the DiffServ model. (2) optimize the use of network throughput and network resources and (3) Adapt to network congestion. 3. It is important to note that QoS routing and resource reservation (discussed in the Section 4) are separate issues. Instead. virtual circuit connection for each ﬂow admitted to the network.
e. CEDAR includes three key components: (1) core extraction. CEDAR proposes the use of core-based routing mechanisms for two primary reasons. Mob.e. Furthermore. Comparison of QoS routing algorithms for ad hoc networks. 2002. local broadcast may be highly unreliable in mobile ad hoc networks. Each node that is not in the core chooses a core neighbor as its dominator. in which a core node tunnels the route request to each of its core neighbors. Core extraction. state propagation and path restoration) is expected to increase network performance and increase network scalability.e. Link state propagation. the core broadcast. reducing the number of nodes participating in route maintenance (i. Typically. experiences a signiﬁcant change (i. CEDAR Ticket-based routing (Bandwidth or delay C link cost Local Bandwidth routing Bandwidth Maintains distance information for every node Periodically Distributed On-demand Flat Yes QoS over AODV Bandwidth or delay Only local.A SURVEY ON QOS SUPPORT FOR MOBILE NETWORKS Table I. routing protocols in ad hoc networks use a broadcast approach to determine routes by using a ﬂooding-based algorithm. Ltd. These protocols differ with regard to the design choices discussed above. First.b). The core of the network is extracted by approximating a minimum dominating set (MDS) of the ad hoc network using only local computation and local state information. 3. is a core node) or is a neighbor of a node in the MDS. because of the hidden terminal and exposed terminal problems. 22]. such that every node in the network is in the MDS (i. Research addressing the QoS routing problem in mobile ad hoc networks has been documented in [8–11]. Each node in the core then establishes a unicast virtual link with other core nodes a distance of three or less away from it in the ad hoc network. Table I presents a summative comparison of the design choices for each QoS routing protocol presented in this section. 2:503– 513 . has been shown to be very unreliable because of the hidden and exposed terminal problems [8. When a link.1. next hop info On-demand as a response to a RREQ message Distributed On-demand Flat No 507 QoS metric State maintenance Bandwidth Local C all high bandwidth stable links When bandwidth changes by some threshold Distributed On-demand Clustered (core nodes required) No QoS state propagation Routing class Route computation Routing architecture Maintains multiple paths Periodically Distributed On-demand Flat Yes routing protocol must also deal with imprecise state information due to node (router) movement and topology changes. Using only a subset of nodes should reduce the negative effects of local broadcast. Copyright 2002 John Wiley & Sons. changes by some threshold value) in available Wirel. To help reduce the effects of these problems. QoS routing in CEDAR is achieved by propagating the bandwidth availability information of stable links in the core subgraph. CEDAR uses the core broadcast mechanism to ﬁnd a route from the dominator of the source to the dominator of the destination. while maintaining low-controloverhead. (a. because of the bandwidth and power constraints. The ﬂooding of route request (RREQ). a QoS routing scheme for ad hoc networks must balance efﬁciency and adaptivity. Comput. Second. The core nodes are responsible for collecting local topology information and performing routing on behalf of the nodes in their respective domain (or immediate neighborhood). The MDS is the minimum subset of nodes. consisting of tens to hundreds of nodes. A Core-Extraction Distributed ad hoc Routing Algorithm The Core-Extraction Distributed ad hoc Routing Algorithm (CEDAR)  has been proposed as a QoS routing algorithm for small to medium size mobile ad hoc networks. CEDAR uses a unicast mechanism. (2) link state propagation and (3) route computation. Commun. however.
e. The dominators are responsible for propagating state information via slow-moving increase waves and fast-moving decrease waves (i. Ticket-based probing is proposed as a general QoS routing approach for MANETs and can handle different QoS constraints (i. CEDAR iteratively tries to ﬁnd a partial route from the source to the domain of the furthest possible intermediate node in the core path. In the case of route failures. established in phase one. thus called bandwidth routing. The proposed bandwidth routing scheme depends on the use of a CDMA  over TDMA  medium access scheme in which the wireless channel is time-slotted.3.508 D. however. but resources are only reserved on the primary path. D. a and b must inform their respective dominators. when an intermediate node receives a probe. source-initiated re computation is used. Commun. messages) to all other core nodes via the core broadcast mechanism. CEDAR initially attempts to re compute an admissible route at the point of failure. The path redundancy scheme establishes multiple routes for the same connection. a probe with more than one ticket is allowed to split into multiple ones. hop-by-hop) multi path QoS routing scheme for ad hoc networks called ticket-based probing is proposed in . After which. Ltd. searched during route discovery. 2002. bandwidth. Instead of randomly selecting the many potential routes to search for an admissible route. either the shortest widest admissible route will be established or a failure is reported. The ﬁrst phase consists of locating the destination node and establishing a core path to the destination. The basic philosophy is that the information about stable links with large available bandwidths can be made known to nodes far away in the network. packet loss and jitter). The second phase consists of ﬁnding a stable route using the core path. Re routing requires that the source node be informed of a path failure. whether the received probe should be split and to which neighbors the probe(s) should be forwarded. 2:503– 513 . CEDAR is an on-demand source routing algorithm and has three key phases. as a directional guide. Route computation. The ﬁnal phase involves two cooperative mechanisms that dynamically restores or re-computes the QoS-based route upon link failures or topology changes in the network. on the basis of its available state information. ticket-based probing does not use a ﬂooding-based route discovery technique. which can satisfy the requested bandwidth. Ticket-based QoS routing solutions for the bandwidth and delay-constrained routing problems were presented in . In this paper. In the second level of redundancy. In the third level of redundancy. delay. Bandwidth Routing A novel QoS routing protocol for QoS support in mobile ad hoc networks is proposed  and is based on the destination sequenced distance vector (DSDV) routing scheme . only one is used as the primary path while the others serve as backup. The routing protocol provides QoS support via separate end-to-end bandwidth calculation and allocation mechanisms. connection request with tighter requirements are issued more tickets). For the highest level of redundancy. Upon the failure of a link. As a long-term solution. Hence. Mob. Comput. it issues a probe (routing message) to the destination. A ticket is the permission to search a single path. while information about dynamic links or low bandwidth links should remain local. The basic goal of the ticket-based probing scheme is to utilize tickets to limit the number of paths Copyright 2002 John Wiley & Sons. multiple paths are selected.e. Similar to CEDAR. but may consist of more (i. HUGHES bandwidth. the entire network is synchronized on a frame and slot basis.e. it decides. resources are reserved along multiple paths. the source initiates the ticketbased algorithm to locate another admissible route. That is. When a source wishes to discover an admissible route to a destination. Eventually. PERKINS AND H. The path bandwidth between a source and destination is deﬁned as the number of free Wirel. 3. each searching a different downstream sub-path. Using only local information about each stable link. threelevel path redundancy and path repairing. the transmission scale is organized as frames (each containing a ﬁxed number of time slots) and a global clock or time-synchronization mechanism is utilized. 3.e. ticket-based probing attempts to search only the best possible routes. The path-repairing mechanism tries avoid the cost of re routing by attempting to repair the route at the point of failure.2. At an intermediate node. we present only the general ticket-based probing scheme. D. A probe is required to carry at least one ticket. ticket-based probing utilizes three mechanisms: path re routing. Ticket-Based Probing A distributed (i. Imprecise state information can be tolerated and multiple paths are searched simultaneously to ﬁnd the most feasible path. resources are reserved along multiple paths and every packet is routed along each path.
which avoids the need for supporting packet encapsulation.e. The Resource ReSerVation Protocol (RSVP)  has been deﬁned as a signaling protocol for the Internet. can be created whenever the RREQ reaches either the destination node. throughput etc. 3. the per-ﬂow service granularity increases the nodeprocessing requirements and limits network scalability. ﬁnally.4. QoS Signaling in Mobile ad hoc Networks QoS signaling (i.) and measuring the delivered QoS (e.e. these two mechanisms have distinct responsibilities. The service mode ﬁeld speciﬁes the level of assurance requested—reservation Wirel. INSIGNIA has several protocol commands: (1) the service mode. incorporating signaling or control data into data packets). After resources (e. throughput etc. and the most recently seen sequence number for the node D. . Commun. the bandwidth routing protocol maintains secondary paths. is the ﬁrst signaling protocol for mobile ad hoc networks.). Flow restoration and adaptation algorithms are responsible for responding to topology changes and changes in available bandwidth. When the primary path fails. Additionally. a minimal set of QoS extensions has been speciﬁed for the RREQ and RREP messages . a mobile host may specify one of two services: A Maximum Delay and Minimum Bandwidth. This problem is NP-complete and thus.g. Copyright 2002 John Wiley & Sons.e. 14]. The speciﬁc extensions for the routing table and control packets (e. However.g. buffer space. which contains the number of hops required to reach the destination node. we present two signaling protocols [13–15] designed for mobile ad hoc networks.g. bandwidth) are allocated via admission control. Quality of Service over AODV The ad hoc On-Demand Distance Vector Routing Protocol (AODV)  has been proposed for besteffort routing in mobile ad hoc networks. INSIGNIA is constructed on the basis of a philosophy of strict separation of routing. setup and maintain virtual connections. Utilizing in-band signaling prevents QoS-related control packets from having to compete with the data packets for access to the channel and can potentially facilitate rapid restoration of QoS requirements along a new path during topology changes. Moreover. adapting and tearing down real-time connections. D. or an intermediate mode with an valid route to the destination. resource reservation) and the QoS routing are closely related and may be coupled or decoupled in QoS architectures. To support fast rerouting during path failures (e. 2:503– 513 4. it must be capable of meeting the QoS constraints. a topological change). it is not suitable for mobile ad hoc networks because the signaling overhead is too high for the mobile host in ad hoc networks. The INSIGNIA signaling protocol uses per-ﬂow service granularity and is responsible for multiple operations: establishing. INSIGNIA INSIGNIA [13. To provide QoS support. Speciﬁcally. delay. Destination nodes actively monitor ongoing ﬂows. Bandwidth calculation requires knowledge of the available bandwidth on each link along the path as well as resolving the scheduling of free slots. Before a node can rebroadcast a RREQ or unicast a RREP to the source.1. which aim to provide minimum bandwidth assurances to real-time applications. to teardown connections in the network. A Route Reply (RREP) packet. Ltd. QoS signaling is used to reserve resources (e. becomes the primary route) and another secondary is discovered.g. Its primary goal is supporting adaptive services. When a route to a new destination is needed. requires a heuristic approach. packet loss. Mob. Comput. RREQ and RREP messages) are outlined in . a node must send an Internet control message protocol (ICMP) QoS LOST message back to the source. inspecting status information (e. the node broadcasts a RREQ packet to ﬁnd a route to the destination.). See  for details of the bandwidth calculation and slot-assignment algorithms. (2) payload type and (3) bandwidth request. INSIGNIA uses in-band signaling (i. Each node that participates in the routeacquisition process places in its routing table the reverse route to the source-node. However. QoS signaling and forwarding functions. restoring. which is based on measured channel capacity and requested bandwidth. Upon detecting that the requested QoS can no longer be maintained. bandwidth etc. the secondary route is used (i. packet loss. 4. 2002. QoS reporting is used to notify source nodes of the current status of ﬂows.A SURVEY ON QOS SUPPORT FOR MOBILE NETWORKS 509 or available time slots between them. In this section. release resources and. respectively. an IP-based QoS framework. Unlike RSVP. These commands are encoded in the IP option ﬁeld. On the other hand.g. the resources are then periodically refreshed by a soft-state mechanism through the reception of data packets. the QoS routing algorithm must ﬁrst select a feasible path. delay.g.
a separate queue for each ﬂow) and the controlled-load service (speciﬁed in the IntServ model). PERKINS AND H. The admission control processing is modiﬁed to deal with bandwidth ranges. Commun. The following extensions and modiﬁcations have been made to the standard RSVP to obtain this new protocol: 1. ‘A measurement speciﬁcation’ (mspec) is added to allow nodes to learn about ‘downstream’ resource bottlenecks. D. being transported and the bandwidth indicator ﬁeld reﬂects the resource availability at intermediate nodes along the path between a source and destination. 3. INSIGNIA is only one component of the QoS architecture and does not deﬁne speciﬁc algorithms but assumes the availability of routing. Using the bandwidth request ﬁeld. While a solution to the hidden-terminal problem is paramount. maximum and minimum bandwidth requirements). 2:503– 513 . data rates). every node must deﬁne a scheduler or queuing discipline that will determine which packet should be sent during the next transmission time slot. random MAC protocols such as Multihop Access Collision Avoidance MACA  focus primarily on solving the well-known hidden-terminal problem. For example. Quality-of-Service MAC Protocols for MANETs Typically. Comput. 6. which carries a ‘sender measurement speciﬁcation’ (smspec) that is used to allow nodes to learn about ‘upstream’ resource bottlenecks. An application programming interface (API) is introduced to handle bandwidth ranges. a QoS-based MAC protocol must also provide a channel access mechanism that considers the QoS constraints of real-time ﬂows. which extends the basic RSVP protocol to support dynamic QoS in mobile ad hoc networks.e. dRSVP uses separate control and data packets (out-of-band signaling). That is.510 D. 4. 5. 4. Copyright 2002 John Wiley & Sons. node movements). D. The separate control packets will contend for access to the transmission channel with the data packets.g. 5. Mob.e. To provide the ﬂexibility needed in dynamic environments such as MANETs. reducing the effective bandwidth. 2. The payload ﬁeld speciﬁes the type of packet—base QoS or enhanced QoS. if a single node in a particular neighborhood wishes to transmit real-time voice data and all other nodes in the same neighborhood Wirel. A bandwidth allocation algorithm is added that divides up available bandwidth among admitted ﬂows. two levels of scheduling are required. since ﬂow state information must be maintained for every ﬂow in each mobile router. HUGHES (RES) mode or best-effort (BE) mode. the goal is simply to reduce the channel access delay of real-time ﬂows. and an additional trafﬁc speciﬁcation (tspec) in PATH messages. packet scheduling and MAC protocols and is transparent to the underlying scheme. Node-level scheduling is a distributed problem in which the all the nodes belonging to a neighborhood must cooperate to determine which nodes has channel access priority. so that they describe ranges of trafﬁc ﬂows.g. 2002. The network makes a commitment to provide service at a point in this range. an application may state its QoS requirements by specifying the minimum level of acceptable service and the maximum level of service that it can utilize. The MACA protocol tackles the hiddenterminal problem by having a sender and receiver exchange a pair of control packets referred to as a requests-to-send (RTS) and clear-to-send (CTS) control . taking into account the desired range for each ﬂow as well as any upstream or downstream bottlenecks for each ﬂow. The Dynamic QoS concept is based on the Dynamic Resource ReSerVation Protocol (dRSVP) protocol. A reservation notiﬁcation message (ResvNotify) is included in this protocol. Ltd. INSIGNIA may face scalability problems similar to RSVP. Since it is not possible to provide upper bounds on channel access delay when using random access mechanisms. This decision can be determined locally by each node. a source can specify its service request (i. Regarding packet-level scheduling. However. An additional ﬂow speciﬁcation (ﬂowspec) in RESV messages. We refer to these as packet-level scheduling and nodelevel scheduling.2. the reservation requests are speciﬁed as a range of values (e. INSIGNIA has some signiﬁcant advantages with respect to the rapid reestablishment of QoS reservations along a new path after network topology changes (i. The Dynamic QoS Concept: dRSVP The Dynamic QoS concept is proposed in  and is a resource reservation–based approach that ﬁts into the IntServ model. To achieve any level of QoS at the medium access layer in ad hoc networks. The dRSVP is a ﬂexible scheme designed to support QoS reservations in mobile ad hoc networks and uses class-based queuing (e.
e. The ﬁrst data packet of the multimedia stream then makes the reservation along the selected path.11 . which can be used to provide different priorities to short control and data packets. Comput.g. These assumptions may not hold in a mobile ad hoc environment. Service Differentiation-Based MAC Protocols Recently. 2002. For subsequent packets belonging to the same real-time ﬂow. Copyright 2002 John Wiley & Sons. Ltd.g. if node A can hear node B. The IEEE 802.) of ad hoc networks. for non real packets. For datagrams (i.2. any node hearing a real-time packet updates its reservation table and defers transmission for the speciﬁed time. The destination conﬁrms the reservation with an ACK. In the remainder of this section. Mob.A SURVEY ON QOS SUPPORT FOR MOBILE NETWORKS 511 only need to transmit best-effort packets. real-time scheduling information is carried (i. piggybacked ) in the headers of data packets and ACK messages. the source node initiates an RTSCTS dialogue (but not for subsequent packets). Commun.11  standard actually speciﬁes the physical and MAC layers of wireless LANs.e. and an ACK is sent to the source. battery power etc. The MAC protocol is called Distributed Coordination Function (DCF) and is suitable for mobile ad hoc networks. MACA/PR The Multihop Access Collision Avoidance with Piggyback Reservation (MACA/PR) medium access protocol uses a reservation-based mechanism to establish a QoS-based connection over a single link in an ad hoc network . The DCF is a Carrier Sense Multiple Access with Collision Avoidance protocol based on the MACA protocol described above. It combines with a RSVP and a QoS routing algorithm to provide end-to-end (i. which try to provide service differentiation in the ad hoc networking environment. However. The RTS-CTS dialogue helps avoid the hidden-terminal problem . source to destination) QoS capabilities in ad hoc networks.11 MAC protocol. these components are called the MACA/PR architecture and are based on CSMA/CA and TDM (e. 5. the nodes with higher priority should not starve best-effort trafﬁc. bandwidth. Generally these protocols can be described as class-based schemes.1. The IEEE 802. we discuss some of the recent work in the area of QoS-based MAC for ad hoc networks. Collectively. To send the ﬁrst packet. the standard has deﬁned 4 types of Inter Frame Spaces. several QoS-based MAC protocols that have been proposed for ad hoc network. non-real-time packets). which are propagated among neighbors. In .e. the authors proposed a modiﬁed backoff algorithm. MACAW was designed to provide fast recovery from the hidden-terminal problem. each with different channel access priorities. . capable of producing several service classes. Wirel. Thus. Nodes may join or leave neighborhoods (or the entire network) without warning.11 standard only provides best-effort service and makes no provisions for QoS support. the basic access scheme for MACA/PR requires a request-to-send/clear-to-send (RTS-CTS) dialogue followed by the transmission of the data packet. 30] and IEEE 802.e. Making the problem more difﬁcult for ad hoc networks is node mobility. 2:503– 513 5. The source node piggybacks the real-time scheduling information for the next packet transmission in the header of its current packet. It is important to note that MACA/PR relies on two underlying assumptions: 1) that real-time packets arrive at constant time intervals and 2) that communication links are symmetric (i.11 standard. the node wishing to transmit real-time data should be given access priority. which contains the scheduling information for the next packet reception. For real-time transmissions. The authors demonstrate their approach by using three service classes. then node B can hear node A and vice versa). Reservation Tables (RTs). A reasonable solution to this problem must also consider the limited resources (e. Additionally. We classify the proposed schemes as either class or priority-based (e. service differentiation) or reservation-based schemes. providing a timeout mechanism for fast recovery in case of packet collisions. Recent work has attempted to provide differentiated service at the MAC layer by manipulating the contention window  associated with the backoff algorithm. Thus. the medium access mechanisms for MACA/PR are equivalent to MACAW [26. Since many of these mechanisms are direct extensions of the distributed MAC scheme in the IEEE 802. we begin our discussion with a brief overview of the IEEE 802. time-slotted bandwidth reservations). are used to avoid the hidden-terminal problem when transmitting real-time packets as opposed to the RTS-CTS dialogue. Once the packet is received by the destination.g. The DCF uses the well-known exponential backoff algorithm to resolve contention between nodes waiting to access the channel.
Commun. Zhang L. Thus. of 5th International Workshop on Mobile Multimedia Communications (MoMuC. A Flexible Quality of Service Model for Mobile Ad-Hoc Networks. 3rd edition. IEEE Journal on Selected Areas in Communication Special Issue on Ad hoc Networks. Summary Mobile ad hoc networks are self-organizing. 3. Sivakumar R. Adaptive Mobile Multimedia Networks. A Framework for QoS Based Routing in the Internet. Liu J-S. adaptive QoS support and robustness) needed for the successful deployment of ad hoc networks are competing and conﬂicting. 19. Resource reSerVation Protocol (RSVP)-Version 1 Functional Speciﬁcation. applications and services equivalent to those available in these environments must be made available to ad hoc network users. QoS routing and MAC). however. Perkins CR. INSIGNIA. Blake S. In Proceedings of IEEE INFOCOM’97. Chua KC. 234– 244. Haas ZJ. In ACM SIGCOMM. Quality of Service for Ad Hoc On-Demand Distance Vector (AODV) Routing. Clark D. Sobrinho JL. 1999. 15. However. We also described the ﬁrst attempt to deﬁne a QoS model for ad hoc networks. Braden R. Many network mechanisms and protocols must cooperate to provide QoS in ad hoc networks. IEEE Personal Communications Magazine 1996. QoS routing in ad hoc wireless networks. because of the limited bandwidth in ad hoc networks. 6(2): 22–30. 7. 6(1): 6–19. Shenker S. Rajagopalan B. any terminal. This paper has presented a broad view of the current research related to providing QoS support in wireless mobile ad hoc networks. Bhagwat P. 98). Chen S. Many of the goals (e. Mirhakkak M. This is the challenge for network researchers and designers. Bharghavan V. Fasbender et al. Nair R. Bambos N. Crawley E. Ltd. 11. QoS provisioning in ad hoc network. 2002. Nair R. 2(1): 15–18. Srivastav M. given that the timescales over which new routes are computed is much faster than those of traditional wired or infrastructured networks. 17. 9. Lin CR. p. Mobile Computing and Communications Review 1997.512 D. A Framework for QoS Based Routing in the Internet. Berson S. In order to achieve this status. 18. INSIGNIA: In-band signaling support for QOS in mobile ad hoc networks. Lee S-B. Lettieri P. 23. decoupling routing and resource reservation may lead to severe performance degradation. Highly dynamic destination sequenced distance vector routing (DSDV) for mobile computers. Germany. IETF RFC2475. anywhere. we have given a detailed discussion of the proposed protocols and algorithms. 5. In IEEE VTC2000-spring. Villasenor J. Internet-Draft. October 1994. RFC 2386. Thus. November 1998. the strict separation of various QoS components (i. IEEE Journal on Special Areas in Communications 1999. 2. Mobile Ad hoc Networking (MANET): Routing Protocol Performance Issues and Evaluation Considerations. 21. Japan. Royer E. Alwan A. Mobile Ad Hoc Networking Terminology. Distributed quality-of-service routing in ad hoc networks. 20. The MITRE Corporation. October 1998. Integrated Services in the Internet Architecture. Internet Draft. 4. IETF RFC1663.g. New York. CEDAR: a coreextraction distributed ad hoc routing algorithm. 17(8): 1488– 1505. 1996. Lee S-B. 1997. routing and resource reservation) may not be appropriate for MANETs. Berlin. Kurose JF. Tanenbaum AS. July 2000. May 2000. A more suitable approach may be the integration of QoS routing and resource management. we believe that designing simple and lightweight (low overhead) protocols should be of primary consideration instead of designing a powerful but more complex QoS system that consumes a large proportion of bandwidth and battery power. 22. Xiao H. Lo A. Gerla M. Macker J. Advances in wireless terminals. 17(8): 1426– 38. Corson S. 12. QoS support is a difﬁcult but vital component to achieving this status. For each component (i. 24. Sandrick H. Rajagopalan B. Copyright 2002 John Wiley & Sons. IEEE JSAC 1999. December 1998. 8. Computer Networking: A Top-down Approach Featuring the Internet. Kleinrock L. rapidly deployable and require no ﬁxed infrastructure [1–3]. Addison Wesley: Reading. Because of bandwidth and power constraints. D. Seah WKG. Crawley E. pp. Braden R. Nahrstedt K. Short J.an Overview. Internet-Draft. 13.txt. 16. August 1998.e. Campbell AT. August 1998. Prentice Hall: Upper Saddle River. MACA/PR: an asynchronous multimedia multi-hop wireless network. 25. Thompson D. IEEE Personal Communications 1999. Panel report on ad-hoc networks. An Architecture for Differentiated Services. minimizing network and QoS overhead is a key objective. RFC 2205. Das SR. New Jersey. March 1999. Bagrodia R. Computer Networks. HUGHES 6. draft-ietf-manet-insignia-01. appropriately balances protocol adaptivity and control overhead). Internet Draft. Ross KW. Perkins C. Furthermore. Jamin S.e. Herzog S. It is hoped that in the future. Perkins CE. ﬂexibility. Wirel. Tokyo. 14. October 1999. Quality-of-service in ad hoc carrier sense multiple access wireless networks. signaling and resource reservation. mobility. 6. Zhang X. Any network. Sinha P. ad hoc networks will emerge as an effective complement to infrastructured wired and wireless LANs and even wide-area mobile networking services. RFC 2386. Dynamic QoS for Mobile Ad Hoc Networks. D. 2001. Mob. 17(8): 1353– 1368. References 1. 3(2): 34– 51. Lin CR. 2:503– 513 . such as PCS. Ahn G-S. 10. September 1997. IEEE Personal Communications 1999.g. MA. it appears that the most appropriate solution. PERKINS AND H. Gerla M. Schultz N. In Proc. will be one that makes the ‘best’ compromise (e. April 2000. Krishnakumar AS. Campbell AT. October 1998. In IEEE Infocom‘99. The QoS algorithms for ad hoc networks must be scalable and capable of efﬁciently utilizing scarce network resources subject to expected high rates of topological change. June 1994. Sandrick H. Comput.
the M. Degree in Computer Science from Michigan State University.S. Authors’ Biographies Dmitri D. 2:503– 513 . Herman D. November 1998.S. Wirel. 30. 27. 28. Mob. He is a member of Pi Mu Epsilon.S. Zhang L. Degree in Chemistry from Tuskegee University. 1959. ARRL. Ltd. The Society for Computer Simulation. 2002. He is currently a PhD student in the Department of Computer Science and Engineering at Michigan State Copyright 2002 John Wiley & Sons.a new channel access method for packet radio. Commun. Professor Hughes serves as director of the Wireless Communications and High-Speed Networking and Performance Research Lab. 1973. In Proc. 26. and Phi Kappa Phi. IEEE INFOCOM’2001. 1994. Perkins received the B. and the PhD Degree in Computer Science from the University of Louisiana at Lafayette. Internet-Draft. 2001. Mutka MW.11. pp. He is currently a Professor of Computer Science at Michigan State University. Demers A. Comput. pp. His research interests include computer system performance analysis & QoS issues related to ATM technology and wireless mobile computer networks. Perkins C. Sigma Xi. Karn P. Kang S. Distributed control algorithms for service differentiation in wireless packet networks. In Int’1 Conference on Computer Communication and Network(ICCCN) 2001. Provisioning service differentiation in ad hoc networks by the modiﬁcation of Backoff Algorithm. October 2001. Veres A. Scottsdale. Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Speciﬁcations. 134–140. IEEE Computer Society. Bharghavan V. MACA. 513 University where he serves as a research member and laboratory manager of the Wireless Communications and High-Speed Networking Laboratory. In Proc. Alaska. His research interests include performance analysis and QoS issues related to the network (routing) and the medium access layers in mobile ad hoc networks. 1995. Shenker S. MACAW: a media access protocol for wireless LANs. Hughes received the B. 1997. 1990. Barry M. of ACM SIGCOMM’94. 29. Dmitri is the recipient of the GE Faculty For the Future Fellowship and a Michigan State University Doctoral Fellowship. Anchorage. the M.A SURVEY ON QOS SUPPORT FOR MOBILE NETWORKS 25. Ad Hoc On Demand Distance Vector (AODV) Routing. Campbell AT. AZ. 1963. In ARR1/CRRL Amateur Radio 9th Computer Networking Conference. Upsilon Pi Epsilon. Degree in Computer Science from Tuskegee University. Royer E. The editors of IEEE 802. 212– 225. IEEE.S Degree in Mathematics from Stillman College.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.