Cluster Based Routing Protocol(CBRP) Functional Specification

类别:软件工程 点击:0 评论:0 推荐:

INTERNET-DRAFT                                           Mingliang Jiang
draft-ietf-manet-cbrp-spec-00.txt          National University of S'pore
                                                              Jinyang Li
                                           National University of S'pore
                                                         Yong Chiang Tay
                                           National University of S'pore
                                                             August 1998


     Cluster Based Routing Protocol(CBRP) Functional Specification

Status of this Memo

   This document is an Internet-Draft.  Internet-Drafts are working
   documents of the Internet Engineering Task Force (IETF), its areas,
   and its working groups.  Note that other groups may also distribute
   working documents as Internet-Drafts.

   Internet-Drafts are draft documents valid for a maximum of six months
   and may be updated, replaced, or obsoleted by other documents at any
   time.  It is inappropriate to use Internet- Drafts as reference
   material or to cite them other than as "work in progress."

   To view the entire list of current Internet-Drafts, please check the
   "1id-abstracts.txt" listing contained in the Internet-Drafts Shadow
   Directories on ftp.is.co.za (Africa), ftp.nordu.net (Northern
   Europe), ftp.nis.garr.it (Southern Europe), munnari.oz.au (Pacific
   Rim), ftp.ietf.org (US East Coast), or ftp.isi.edu (US West Coast).


Abstract

   Cluster Based Routing Protocol (CBRP) is a routing protocol designed
   for use in mobile ad hoc networks.  The protocol divides the nodes of
   the ad hoc network into a number of overlapping or disjoint clusters
   in a distributed manner.  A cluster head is elected for each cluster
   to maintain cluster membership information. Inter-cluster routes are
   discovered dynamically using the cluster membership information kept
   at each cluster head.  By clustering nodes into groups, the protocol
   efficiently minimizes the flooding traffic during route discovery and
   speeds up this process as well.  Furthermore, the protocol takes into
   consideration of the existence of uni-directional links and uses
   these links for both intra-cluster and inter-cluster routing.


Jiang, Li, Tay                                                  [Page 1]

INTERNET-DRAFT        CBRP Functional Specification          August 1998


1 Introduction

   There are several major difficulties for designing a routing protocol
   for MANET.  Firstly and most importantly, MANET has a dynamically
   changing topology due to the movement of mobile nodes which favors
   routing protocols that dynamically discover routes over conventional
   distant vector routing protocols [6], e.g. Dynamic Source Routing[4],
   TORA[3], ABR[5] etc.  Secondly, the fact that MANET lacks any
   structure makes IP subnetting impossible.  However, routing protocols
   that are flat, i.e. only one level of hierarchy, might suffer from
   excessive overhead when scaled up.  Thirdly, links in mobile networks
   could be asymmetric at times. Routing protocols could make use of the
   uni-directional links to improve its overall performance.

   CBRP has the following features:

   * fully distributed operation.

   * less flooding traffic during the dynamic route discovery process.

   * explicit exploitation of uni-directional links that would otherwise
     be unused.

   The idea of using clusters for routing in Ad hoc networks has
   appeared in [7],[8].   In these protocols clusters are introduced to
   minimize updating overhead during topology change. However, the
   overhead for letting each and every node maintain up-to-date
   information about the whole network's cluster membership and inter-
   cluster routing information in order to route a packet is
   considerable.  In comparison, simpler and smaller clusters are found
   in [9] and [10], however, the use of these clusters is mainly for the
   task of channel assignment; how they can help in the routing process
   is not discussed.

   CBRP adopts the cluster formation algorithm as proposed in [9], but
   unlike [9], CBRP mainly concentrates on the use of clusters in the
   routing process.  CBRP is different from previously proposed cluster-
   based routing algorithms and it uses a dynamic route discovery
   strategy.

2. CBRP terminology

   This section defines terms used in CBRP that do not appear in [2]:

   * Node ID

   Node ID is a string that uniquely identifies a particular mobile
   node.  Node IDs must be totally ordered. In CBRP, we use a node's IP

Jiang, Li, Tay                                                  [Page 2]

INTERNET-DRAFT        CBRP Functional Specification          August 1998


   address as its ID for purposes of routing and interoperability with
   fixed networks.

   * Cluster

   A cluster consists of a group of nodes with one of them elected as a
   cluster head. The election procedure is described in section 5.1 and
   5.2.  A cluster is identified by its Cluster Head ID. Clusters are
   either overlapping or disjoint.  Each node in the network knows its
   corresponding Cluster Head(s) and therefore knows which cluster(s) it
   belongs to.

   * Host Cluster

   A node regards itself as in cluster X if it has a bi-directional link
   to the head of cluster X.   In such a case, cluster X is a host
   cluster for this node.

   * Cluster Head

   A cluster head is elected in the cluster formation process for each
   cluster.  Each cluster should have one and only one cluster head.  A
   cluster head has complete knowledge about group membership and link
   state information in the cluster  within a bounded time once the
   topology within a cluster stabalizes.  In CBRP, each node in the
   cluster has a bi-directional link to the cluster head.

   * Cluster Member

   All nodes within a cluster EXCEPT the cluster head are called members
   of this cluster.

   * Gateway Node

   A node is called a gateway node of a cluster if it KNOWS that it has
   a bi-directional or uni-directional link to a node from another
   cluster. As shown in Figure 1, node 3 and 4 are gateway nodes of
   cluster 1, node 6 is a gateway node of cluster 8.

           2       5
         //         \\
        //           \\
       (1)===3==6====(8)
        \\           //
         \\         //
           4<------7

             Fig. 1

Jiang, Li, Tay                                                  [Page 3]

INTERNET-DRAFT        CBRP Functional Specification          August 1998


   * Neighbor Table

   The neighbor table is a conceptual data structure that we employ for
   link status sensing and cluster formation. Each entry contains the ID
   of a neighbor that it has connectivity with and the status of that
   link and the role of the neighbor (a leader or a member).


3. Physical and Link Layer Assumptions

   This section lists the assumptions we made about the underlying
   physical and link layers when designing CBRP.  Each node in MANET is
   equipped with a transceiver.  In general, CBRP works best with
   omnidirectional antennas. Each packet that a node sends is broadcast
   into the region of its radio coverage.


4.   Link/Connection Status Sensing Mechanism

   In CBRP, each node knows its bi-directional links to its neighbors as
   well as unidirectional links from its neighbors to itself.  For this
   purpose, each node maintains a Neighbor Table as follows:

   +------------+-------------------------------+---------------------+
   | NEIGHBOR_ID| LINK_STATUS                   | ROLE                |
   +------------+-------------------------------+---------------------+
   | neighbor 1 | bi/unidirectional link to me? | is 1 a cluster head?|
   +------------+-------------------------------+---------------------+
   | neighbor 2 | bi/unidirectional link to me? | is 2 a cluster head?|
   +------------+-------------------------------+---------------------+
   |  ...
   +------------+-------------------------------+---------------------+
   | neighbor N | bi/unidirectional link to me? | is N a cluster head?|
   +------------+-------------------------------+---------------------+

   Each node broadcasts its Neighbor Table periodically in a HELLO
   message as shown below every HELLO_INTERVAL.

   +-----------+------------------------------------------------------+
   | MY_OWN_ID | MY_MEMBERSHIP_STATUS                                 |
   |           | (CLUSTER_HEAD/CLUSTER_MEMBER/UNDECIDED)              |
   +-----------+------------------------------------------------------+
   | Neighbor Table                                                   |
   +------------------------------------------------------------------+

   The first field specifies the ID of the sender.  The second field
   tells whether the sender is a cluster head, cluster member or
   undecided.  ("undecided" means a node is still in search of its host

Jiang, Li, Tay                                                  [Page 4]

INTERNET-DRAFT        CBRP Functional Specification          August 1998


   cluster)

   Upon receiving a HELLO message from its neighbor B, node A modifies
   its own Neighbor Table as follows:

       1. It checks if B is already in the Neighbor Table, if not, it
          adds one entry for B.
       2. If B's Neighbor Table contains A, A marks the link to B as
          bi-directional in the relevant entry.
       3. If B is cluster head, marks B as a cluster head in the entry.

   Each entry in the Neighbor Table is associated with a timer. A table
   entry will be removed if a HELLO message from the entry's node is not
   received for a period of (HELLO_LOSS+1)*HELLO_INTERVAL, allowing
   HELLO_LOSS consecutive HELLO messages to be lost from that node.

   When a nodes' neighborhood topology stabilizes, each node will have
   complete knowledge of all the nodes that it has a bi-directional link
   to and all the nodes that have a uni-directional link to it within a
   bounded time.  However, a node would not know to whom it has a uni-
   directional link.

   Note that the 2nd and 3rd column of the neighbor table need not be
   sent out for the purposes of link sensing, however, they are there
   mainly for cluster formation and other functionality of the protocol.


5. Protocol Operation


   The operations of CBRP are entirely distributed.  According to the
   functionality, CBRP could be viewed as a combination of the three
   following sub-functions which will be discussed below.

5.1 Cluster Formation

   The Cluster Formation algorithm is a simple "lowest ID" clustering
   algorithm in which the node with a lowest ID is elected as the
   Cluster Head [9].

   A node uses the information obtained from the HELLO message for
   Cluster Formation.  A node that has the lowest ID among all its bi-
   directionally linked neighbors (a node that has given up the role as
   a Cluster Head is not counted, i.e. a cluster member node ). The new
   Cluster Head will change the first field in its subsequently
   broadcast HELLO messages from "undecided" to "cluster head"
   thereafter.


Jiang, Li, Tay                                                  [Page 5]

INTERNET-DRAFT        CBRP Functional Specification          August 1998


   A cluster head regards all the nodes it has bi-directional links to
   as its member nodes. A node regards itself as a member node for a
   particular cluster if it has a bi-directional link to the
   corresponding cluster head.  Note that a member node may hear from
   several cluster heads and therefore have several host clusters.

5.2 Cluster Maintenance

   As clusters are identified by their respective cluster heads, we
   would like to have the cluster heads change as infrequently as
   possible.  We use the following rules for changing cluster head, as
   described in [10].

     1. A non-cluster head never challenges the status of an existing
        cluster head, i.e. if X is a non-cluster head node with a
        bi-directional link to cluster head Y, X does not become
        a cluster head even if it has an ID lower than Y's.
     2. Only when two cluster heads move next to each other (i.e. there
        is a bi-directional link between them), one of them loses the
        role of cluster head.

   Therefore, our exact algorithm is not a strict "lowest ID" clustering
   algorithm.  The cluster maintenance procedure are discussed under
   three subsections:

   * Node Removal

   A node X is removed from a host cluster either because it loses the
   bi-directional links to the cluster head, or because of node
   failures.  In either case, the cluster head and node X will time out
   the corresponding entries in their Neighbor Table so that the cluster
   information will be updated.

   * Node Addition

   A member node is added to a new cluster when it discovers a bi-
   directional link to the respective cluster head even if the cluster
   head has a higher ID, which is consistent with rule 1.  The node will
   know its new host cluster and the new host cluster head will know its
   new member through the updated Neighbor Table.

   When a node is switched on, its MY_MEMBERSHIP_STATUS field in the
   HELLO message is initially UNDECIDED for a period of UNDECIDED_PD. If
   it discovers that it has a bi-directional link to a cluster head, it
   modifies this field as CLUSTER_MEMBER. If there is no such bi-
   directional links to any cluster head, it takes the role as a cluster
   head and indicates this in the subsequent HELLO messages it sends.


Jiang, Li, Tay                                                  [Page 6]

INTERNET-DRAFT        CBRP Functional Specification          August 1998


   * Cluster Head Contention

   When two Cluster Heads move next to each other (till they have a bi-
   directional link in-between), one of them must give up its role as
   cluster Head.  As a result, whenever a cluster head hears the HELLO
   message from another cluster head indicating a bi-directional link in
   between, it will compare its own ID with that of the other cluster
   head's. The one with a smaller ID will continue to act as cluster
   head.  The one with a bigger ID gives up its role as cluster head and
   changes from "cluster head" to "member" in its subsequent HELLO
   messages. This might trigger the formation of other new clusters.

5.3  Routing Considerations


   Routing in the CBRP is based on source routing, which is similar to
   [4]. Therefore, it can be viewed as consisting of 3 phases: route
   discovery, packets routing and stale route removal, of which packet
   routing is trivial.

   Cluster structure is exploited to minimize the flooding traffic
   during route discovery phase. Furthermore, some uni-directional links
   are discovered and used which increases network connectivity. This is
   discussed in the Gateway Discovery section.


5.3.1 Gateway Discovery

   Cluster X and cluster Y are said to be bi-directionally linked, if
   any node in cluster X is bi-directionally linked to another node in
   cluster Y, or if there is a pair of opposite uni-directional links
   between any 2 nodes in cluster X and cluster Y respectively.

   Cluster X is said to be uni-directionally linked to cluster Y if any
   node in cluster X is uni-directionally linked to any other node in
   cluster Y. Y is called X's upstream uni-directionally linked
   neighboring cluster.

   By Gateway Disvoery, a cluster head for cluster X will obtain the
   information about cluster X's bi-directionally and upstream uni-
   directionally linked neighboring clusters.

   For this purpose, a Cluster Adjacency table is kept at each node
   which is formatted as follows:

Jiang, Li, Tay                                                  [Page 7]

INTERNET-DRAFT        CBRP Functional Specification          August 1998


   +------------------+-----------------------------------------------+
   |Adjacent Cluster1 | adjacent node|to/from/bi                      |
   +------------------+-----------------------------------------------+
   |Adjacent Cluster2 | adjacent node|to/from/bi                      |
   +------------------+-----------------------------------------------+
   |...                                                               |
   +------------------------------------------------------------------+

   This table is updated by the periodic HELLO message a node hears.
   Note that a node may have several links to nodes in a particular
   cluster and only one of them is recorded in the Cluster Adjacency
   Table.  The selection rule is as follows:

      1. A bi-directional link takes precedence over all uni-directional
         links.
      2. Of the links that have the same precedence, the one with the
         lowest ID wins.

   This table is periodically sent to the member node's host cluster
   heads.  (It could be piggybacked to the HELLO message if possible.) A
   cluster head uses its members' Cluster Adjacency table to construct
   its own cluster adjacency table.  The construction rule is the same
   as that of the member's.

   Cluster head will flood its neighboring clusters with a message of
   TTL 3 in search for the "to" link that corresponds to a "from" link
   in its Cluster Adjacency Table. As a result, cluster heads will have
   complete knowledge of all its bi-directionally linked neighboring
   clusters even if there is no actual bi-directional links in between.
   For example, in Fig 1, Cluster 1 knows that 2 can reach cluster 1
   through 5, but 1 does not know that cluster 2 can be reached by node
   3. In CBRP, cluster head 1 and 2 will discover this scenario and
   disseminate the information to node 3 and 5 respectively.

            3-->4
          //     \\              1 and 2 are heads of their respective
         //       \\             clusters, 3,4,5,6 are members.
        (1)       (2)
         \\       //
          \\     //
            6<--5

            Fig. 2


Jiang, Li, Tay                                                  [Page 8]

INTERNET-DRAFT        CBRP Functional Specification          August 1998


5.3.2 Route Discovery


   Route Discovery is the mechanism whereby a node S wishing to send a
   packet to a destination D obtains a source route to D. Similar to
   other MANET protocols [4] [1], the way S finds a route(or multiple
   routes) to D is also done by flooding, however, because of the
   clustering approach the number of nodes that are disturbed are much
   less in general.

   Essentially in Route Discovery, cluster heads are flooded in search
   for a source route. To perform Route Discovery, the source node S
   sends out a Route Request Packet (RREQ), with a recorded source route
   listing only itself initially. Any node that forwards this packet
   will append its own ID in this RREQ.  Each node forwards a RREQ
   packet only once and it never forwards it to a node that has already
   appeared in the recorded route. In CBRP, the RREQ will always follow
   a route with the following pattern to reach destination D:
           S,CH1,G1,CH2,G2,G3,CH3 ..... D

   A detailed description of how this is achieved is presented below.
   Source always unicasts RREQ to its cluster head, say CH1.  Each
   cluster head will unicast RREQ to each of its bi-directionally linked
   neighbors which have not yet appeared in the recorded route through
   the corresponding gateway. This process continues until the target is
   found or until another node that can supply a route to the target is
   found.


   When the target of the Request, node D, receives the RREQ, D may
   choose to memorize the reversed source route to S.  Node D then
   copies the recorded source route into a Route Reply packet(RREP)
   which it then sends back to the initiator of the Route Request (e.g.,
   node S) by reversing the recorded route and putting it in the IP
   header of the Route Reply packet. The recorded route gives the
   complete information about the SEQUENCE OF CLUSTERS source should
   traverse in order to reach destination D.  While forwarding the Route
   Reply, intermediate cluster heads modifies the IP header of the
   packets, substitute the inter-cluster incoming links to inter-cluster
   outgoing links. Each intermediate cluster head also modifies the
   recorded route in the Route Reply packet to optimize the recorded
   route as much as possible using its knowledge of the cluster topology
   and inter-cluster gateway information. An example of such
   optimization is to connect two gateway nodes by an intra-cluster link
   that does not go through the cluster head.

   All source routes learned by a node are kept in a Route Cache, which
   is used to further reduce the cost of Route Discovery.  When a node

Jiang, Li, Tay                                                  [Page 9]

INTERNET-DRAFT        CBRP Functional Specification          August 1998


   wishes to send a packet, it examines its own Route Cache and performs
   Route Discovery only if no suitable source route is found in its
   cache.


5.3.3 Route Removal

   A source route is removed from S if the network topology has changed
   such that S can no longer use the route to D because a hop along the
   route no longer works. If S still wants to communicate with D, it can
   invoke Route Discovery again to find a new route.

6. Discussions and Implementation Considerations

   This section discusses some of the problems that we have encountered
   during the design of CBRP. In particular, they are related to the use
   of uni-directional links in routing.

6.1 ARP and Uni-directional links

   In a MANET context, links between 2 nodes can be bi-directional and
   uni-directional.   However, when the link layer does MAC-layer
   address based packet filtering, (which most current technologies do),
   special care has to be taken for ARP for uni-directional links.  For
   example, when there is a uni-directional link from node A to node B
   as shown in Figure 2, node A should not rely on the conventional ARP
   protocol to resolve node B's MAC layer address, because node B's ARP
   reply will never reach node A directly.

          A---->B

          Fig. 3


   CBRP is able to use uni-directional links. For an intra-cluster uni-
   directional link, the upstream node can be informed by its cluster
   head of the MAC-layer address of the downstream node. For a inter-
   cluster uni-directional link, as shown in Figure 1, node 3(5) will
   know node 4(6)'s MAC-layer address by the process of gateway
   discovery.

6.2 Rate of Stale Route Discovery and Uni-directional Links

   In general (not pertaining to CBRP),  when uni-directional links are
   used, discovery of stale routes can be slow. As shown in Figure 3,
   when the link between 1 and 2 breaks, node 1 will not be aware until
   a message comes from 2 by way of 3,4,5,6. This observation justifies
   CBRP's use of inter-cluster uni-directional links only between 2

Jiang, Li, Tay                                                 [Page 10]

INTERNET-DRAFT        CBRP Functional Specification          August 1998


   clusters.

              1--->2--->3--->4--->5
              ^                   |
              |                   |
              |                   |
              |                   |
              +-------- 6 <-------+

                     Fig. 4

7. Future Directions

   Cluster based routing protocols (CBRP) is the first step towards a
   more scalable solution using clustering approach. It also suggests
   how uni-directional links in Ad hoc networks can be used in routing
   and the reveals problems resulting from such use.

   Several questions remain to be answered in CBRP,

     1. How collisions can be avoided or handled effectively. Shall we
        have spatial reuse of the channels among different clusters?
     2. Should clusters have native support for QoS as in [10]?
     3. Should clusters be made bigger than two hops diameter? If so,
        what criteria should be used to form a bigger cluster? Or,
        should we use a hierarchy of clusters?  Will the resulted more
        complex cluster formation and maintenance procedure offset the
        advantage of having a bigger size?

   To give satisfactory answers to these interesting questions will be
   our future directions in refining CBRP.


Jiang, Li, Tay                                                 [Page 11]

NTERNET-DRAFT        CBRP Functional Specification          August 1998


References


   [1] Perkins, C.E., "Ad-Hoc On-Demand Distance Vector Routing",
       MILCOM'97 panel on Ad-Hoc Networks, Monterey, CA, November 3,
       1997.

   [2] Perkins, C.E., "Mobile Ad Hoc Networking Terminology",
       draft-ietf-manet-term-00.txt, work in progress.

   [3] Corson, M.S., and Ephremides, A.,  "A Distributed Routing
       Algorithm for Mobile Wireless Networks", ACM/Baltzer Journal
       on Wireless Networks, January 1995.

   [4] Johnson, D.B., and Maltz, D.A., "Dynamic Source Routing in
       Ad-Hoc Wireless Networks", Mobile Computing, T.Imielinski and
       H.Korth, editors, Kluwer, 1996.

   [5] Toh,C.K., "A Novel Distributed Routing Protocol To Support
       Ad-Hoc Mobile Computing", International Phoenix Conference on
       Computers and Communications (IPCCC'96), pg 480-486, March 1996.

   [6] Hedrick, C., "Routing Information Protocol", RFC 1058, June 1998.

   [7] Das, B., Raghupathy, S, Vaduvur, B, "Routing in Ad Hoc Networks
       Using a Spine", Proceedings of 6th International Conference on
       Computer Communications and Networks, Las Vegas, USA, September,
       1997.

   [8] Krishna, P., Vaidya, N.H., Chatterjee, M., Pradhan, D.K., " A
       Cluster-based Approach for Routing in Dynamic Networks",
       Proceedings of the Second USENIX Symposium on Mobile and
       Location-Independent Computing, 1995.

   [9] Gerla, M., Tsai, T.C., "Multiculster, mobile, multimedia radio
       network", ACM/Balzer Journal of Wireless Networks, 1995.

   [10] Chiang, C.C., Wu, H.K., Liu, W., Gerla, M., "Routing in
        Clustered Multihop, Mobile Wireless Networks with Fading
        Channel", The Next Millennium, the IEEE SICON, 1997.


   This work was supported in part by National University of Singapore
   ARF grant RP960683.

Jiang, Li, Tay                                                 [Page 12]

INTERNET-DRAFT        CBRP Functional Specification          August 1998


Authors' Information

   Jiang Mingliang
   Mobile Computing Lab
   School of Computing
   National University of Singapore
   Singapore 119260
   Email: [email protected]

   Li Jinyang
   Mobile Computing Lab
   School of Computing
   National University of Singapore
   Singapore 119260
   Email: [email protected]

   Tay Yong Chiang
   Department of Mathematics
   National University of Singapore
   Singapore 11920
   Email: [email protected]
Jiang, Li, Tay                                                 [Page 13]


本文地址:http://com.8s8s.com/it/it32187.htm