📄 draft-ietf-manet-cbrp-spec-01.txt
字号:
One possible extension to this technology is to allow ACKs to be for- warded by neighboring cluster heads back to the sender if a uni- directional link between two clusters are being used. Such a return path is bounded by maximum 5 hops. The forwarding of ACK could be viewed as collapsing layer 2 and layer 3 routing functionality, a violation of the layering model which states that lower layer should hide details away from upper layers. Although this seems rather inefficient at first sight, if we consider the fact the bi-direc- tional links are preferred over uni-directional links in CBRP, we would realize that such a uni-directional link will not be used unless one cannot reach a specific node using bi-directional links only.7.3 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 4, 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 clusters. 1--->2--->3--->4--->5 ^ | | | | | | | +-------- 6 <-------+ Fig. 48. Constants and Suggested ValuesJiang, Li, Tay [Page 21]INTERNET-DRAFT CBRP Functional Specification July 1999 These are the CBRP constant values that we have experiemented with in the simulation. HELLO_INTEVAL 1.5s or 2s HELLO_LOSS 1 CONTENTION_PERIOD 1.5s9. Applicability Statement This section summarizes the characteristics of CBRP, as specified in the Applicability Statement draft.9.1 Network Context The protocol is designed for medium to large networks with relatively large average nodal degree (> 5). Nodal mobility should not be too high, i.e. node cannot move quicker than the speed of cluster forma- tion which is defined as 1/HELLO_INTERVAL. This protocol is suited for networks where most of the traffic is among a small set of sender-receiver pairs compared to the possibil- ity of N*(N-1)/2 number of pairs. Also, the applications supported should be able to tolerate the delay of route discovery time.9.2 Protocol Characteristics and Mechanisms * Does the protocol provide support for unidirectional links? (if so, how?) Yes. It selectively makes use of those uni-directional links that could give two-way-route to nodes that are otherwise inaccessible using only bi-directional links. * Does the protocol require the use of tunneling? (if so, how?) No. * Does the protocol require using some form of source routing? (if so, how?) Yes. routes are discovered in route discovery stage and will be carried in the packet headers in actual routing. (Refer to Section 6.2) However, source routing is actually notJiang, Li, Tay [Page 22]INTERNET-DRAFT CBRP Functional Specification July 1999 essential to the correct working of CBRP. As source routing poses a non-negligible overhead when the network sizes grow, we are currently designing alternatives to replace it with more efficient mechanisms. * Does the protocol require the use of periodic messaging? (if so, how?) Yes. Periodically, each node in the network sends a HELLO mes- sage containing its current neighbor table. The size of the message is proportional to the degree of the node (i.e. the number of neighbors), which is around 6 to 15 for networks of average density. * Does the protocol require the use of reliable or sequenced packet delivery? (if so, how?) No. * Does the protocol provide support for routing through a multi- technology routing fabric? (if so, how?) Yes. Each network interface is assigned a unique IP address used for routing purpose. * Does the protocol provide support for multiple hosts per router? (if so, how?) Yes. A number of hosts, each having a unique IP address, could associate itself with a router that forwards packets and participates in the routing protocol on the hosts' behalf. * Does the protocol require link or neighbor status sensing (if so, how?) Yes. Neighbor status sensing is required. * Does the protocol have dependence on a central entity? (if so, how?) No. All the functions are achieved in a distributed manner. The cluster formation process differentiates the roles ofJiang, Li, Tay [Page 23]INTERNET-DRAFT CBRP Functional Specification July 1999 nodes as cluster heads and member nodes. Cluster heads are flooded during the route discovery phase to find a route to the destination. However the actual routing will try to bypass cluster heads as intermediate nodes. * Does the protocol function reactively? (if so, how?) Yes. It defers getting the route information until such a route is explicitly asked for by the application. Routing is done on demand with 3 phases: route discovery, packet routing, route removal. * Does the protocol function proactively? (if so, how?) No. But it proactively acquires its 2-hop topology informa- tion through the exchange of HELLO messages. * Does the protocol provide loop-free routing? (if so, how?) Yes. Source routing records all nodes along the route that could be easily checked for loop-freedom. * Does the protocol provide for sleep period operation? (if so, how?) No. * Does the protocol provide some form of security? (if so, how?) Not by itself. It has to rely on other protocols (e.g. IMEP, IPsec) to give security support. * Does the protocol provide support for utilizing multi-channel, link-layer technologies? (if so, how?) Yes. Cluster-formation algorithms could be run on different channels to yield different sets of clusters for each channel. Routing could be done independently on each of the set of clusters.ReferencesJiang, Li, Tay [Page 24]INTERNET-DRAFT CBRP Functional Specification July 1999 [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] Park V., Corson M.S, "A Highly Adaptive Distributed Routing Algorithm for Mobile Wireless Networks," Proc. IEEE INFOCOM '97, Kobe, Japan, Apr 1997. [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. [11] Ephremides A., Wieselthier J.E., Baker D.J., "A Design Concept for Reliable Mobile Radio Networks with Frequency Hopping Signaling", Proceedings of the IEEE, Vol.75, No.1, Jan 1987Jiang, Li, Tay [Page 25]INTERNET-DRAFT CBRP Functional Specification July 1999Appendix A +---------------------------> C_UNDECIDED | +-------------------+------------------------+ | | | | | +-----+----------+ +----+----------+ +------+--------+ | / receive HELLO / / u_timer times / / receive HELLO / | / from non-head / / out / / from head / | +---------+------+ +--------+------+ +----------+----+ | | | | | +----------+----------+ (Neighbor Table +--------+--------+ | |update Neighbor Table| Empty?) |kill u_timer | | +----------+----------+ YES/ \NO |update Neighbor | | | +------------+ +----+----------+ |Table as C_MEMBER| +-------------+-+schedule new| |send triggered | +--------+--------+ | |u_timer | |HELLO as C_HEAD| | | +------------+ +---------+-----+ | | | | | +--------------------> C_HEAD <--------+ | | | +----------------+----------------+ +------+ | | | | | | | | +-------------+ +--------------+ +--------+ | | | /receive HELLO/ / receive HELLO/ /c_timer / | | | /from non-head/ / from C_HEAD / / expires/ | | |+-----+-------+ +--------+-----+ +-----+--+ | | | | | | | | | | +-----------------+(Still in contention | | |+---------------+ |receive HELLO | with the other head?) | | ||update Neighbor| |from C_HEAD,set | NO/\ YES | | ||Table | |c_timer | / (my ID smaller?) | | |+-----+---------+ +-------+---------+ / / \ NO | | | | | /YES/ +----+------------+ | | +------+-------------------+---------+----+ |send triggered +-+ | | |HELLO as C_MEMBER| | | | +-----------------+ | | | C_MEMBER <---------------------------+ | | +-------+----------------------+ | | | | | | | | +--------+-----+ +------+-------+ | | | /last head lost/ / receive HELLO/ | | | +----------+---+ +--------+-----+ | | +---------------+ | | | | |send triggered +-YES-+(my ID smallest?) +--------+-------+ | | |HELLO as C_HEAD| | | update Neighbor+--+ | +---------------+ NO| | Table | | +-------+--------+ +--------+-------+ +----------------------|schedule u_timer| +----------------+Jiang, Li, Tay [Page 26]INTERNET-DRAFT CBRP Functional Specification July 1999 This work was supported in part by National University of Singapore ARF grant RP960683.Authors' Information CBRP's URL: Contains information on CBRP's latest development and simulation code http://cram.comp.nus.edu.sg:8080/cram/cbrp/ Jiang Mingliang Mobile Computing Lab School of Computing National University of Singapore Singapore 119260 Email: jiang@eudoramail.com Li Jinyang Mobile Computing Lab School of Computing National University of Singapore Singapore 119260 Email: lijy@acm.org Y.C. Tay Department of Mathematics National University of Singapore Singapore 119260 Email: tay@acm.orgJiang, Li, Tay [Page 27]
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -