Performance Enhancement of Ad Hoc Networks with Adaptive Monitor Based Routing Tanay Dey, Subrota Kumar Mondal and M.M.A Hashem Deptartment of Computer Science and Engineering Khulna University of Engineering and Technology (KUET) Khulna-9203, Bangladesh
[email protected], subrota
[email protected], mma
[email protected] the data are destined with purity. As routing overhead was drastically cut down, AMBR without any doubt does not consume limited network bandwidth. In section II, we describe our proposed protocol AMBR and in section III, we describe the analytical model of the proposed protocol. Simulation results and performance comparison are shown in section IV. Finally, section V concludes the paper and outlines the future work.
Abstract— To effectively support the communication in such a dynamic networking environment as the ad hoc networks, the routing protocol has to be adaptable to the spatial and temporal changes in the characteristics of the network, such as traffic and mobility patterns. In this paper, we propose and analyze a routing protocol called Adaptive Monitor Based Routing (AMBR) that discovers and maintains routes in hierarchical and distributed fashion and locally repairs the broken link. The main motivation behind the proposed AMBR is to drastically cut down flooding, to substantially tame the routing overhead, to repair broken link locally in order to minimize the routing overhead and to increase efficiencies in packet movement in the ad hoc networks. An analytical model for the AMBR technique is presented. Simulation results illustrate its performance and demonstrate its good behavior comparing with other prevalent protocols. The proposed protocol achieved substantial improvement in terms of flooding, routing overhead and network overhead and hence, provides enhanced reliability. Index Terms— Ad hoc network, adaptive routing, dynamic broken link, monitor, routing control overhead.
II. P ROPOSED AMBR ROUTING P ROTOCOL Our proposed protocol AMBR uses nodes which are called monitors, whose first hop connectivity (total no of neighbors) acquire a predefined number over the whole network. If a node gets alive, it broadcasts a message named ”hello”. This ”hello” message is not periodic, rather it is event driven. Any node, getting this message from another node should give back a reply back to that node only (not a broadcast). Detecting of monitor is completely individual responsibility and it is done in every node getting the information of the network. If a node has calculated that it’s total number of neighbors or first hop connectivity is equal to a predefined number then it will broadcast a ”new monitor” packet and all its covered nodes will accept it and the ”hello” propagation immediately ends as it is no longer an ordinary nodes, it has now become a monitor. In the case of rapid network change, number of neighbors increase or decrease of a monitor, then the monitor does not need to inform its present neighbors about this change. The purpose of a network is to send or receive data so that the monitor available messages can be piggy backed with any data or acknowledge packet that has gone to monitor from that node or vice versa. Any node that has not sent any data to monitor and has not got any data from monitor over a predefined time, then to avoid complexity, every node must inform its monitor that it is in its zone by a periodic control packet named ”monitor alive request”. This message does not continue in the ”hello” session, in fact, when a ”hello” session starts it stops and starts when the ”hello” session stops with the information given by the ”hello” session. The monitor will reply this request with the ”monitor alive request repeat” packet only
I. I NTRODUCTION An ad hoc network is a class of wireless systems that consists of independent mobile nodes communicating with each other over wireless links, without any static infrastructure such as base stations [1]-[4]. As ad hoc networks do not rely on existing infrastructure and are self organizing, they can be rapidly deployed to provide robust communication in a variety of hostile environments. It is not difficult to envision ad hoc networks operating over a wide range of coverage areas, node densities, mobility patterns and traffic behaviors [5]. In CSGR [6] problem with having explicit cluster heads is that routing through cluster heads creates traffic bottlenecks. In Landmark, LANMAR [8] and L+ [12], this is partially solved by allowing nearby nodes route packets instead of the cluster head. In this paper, a distributed and hierarchical routing protocol has been proposed based on first hop connectivity of every mobile node existing in the ad hoc networks. The route discovery algorithm for AMBR is kept as simple as possible but is made a powerful algorithm to find the path to destination, as well as maintaining the path until
1-4244-1463-6/08/$25.00 © 2008 IEEE
263
RWS 2008
Here, (see Fig. 3) the monitor getting a request from source node finds that it is not directly connected to the destination. So, it searches the cache and getting the route it delivers data towards that way.
to the node from where it received the request. If the monitor does not receive piggy backed information within a predefined amount of time and no active data delivery is in that session or no monitor alive request packet from a neighbor node in that session then it remove that node from the neighbor node list and if the total number of neighbors is less than the predefined number then it starts a new ”hello” session. In this approach, for the data transfer there are couples of cases that are needed to be considered: 1) If the destination (D) is directly connected with the source (S) (see Fig. 1) then the data is simply sent to destination
D M
S
Fig. 3.
D
Monitor has the destination in its cache
c) If the source node finds that the route to the destination is in its cache then it follows the route (see Fig. 4) but if the route gets broken dynamically, the monitor currently has the packet, propagates queries to all the directly connected monitors of it about the destination (see Fig. 5). This process continues upwards. If any of them has it, then they may follow the reverse path to reply the query made by the monitor. If multiple paths are found then the monitor takes the path from which the reply came first, it keeps that path in its cache. In this way protocol recover the dynamically broken link.
S
Fig. 1.
M
Source is directly connected to destination
2) If the destination is not directly connected then the following cases may appear: a) If the source (S) finds that the destination is not directly connected to it and the route of the destination is not in the cache then it sends the data to its monitor (M ) and its monitor on behalf of it finds the desired destination. If the destination is directly connected to the monitor then it sends the data on behalf of the source (see Fig. 2).
D
M
M
M
M
M M
M
S
D
M S Fig. 4.
Source has the destination in its cache
d) If the route wanted by the source does not evolve in the route cache of the monitor, it propagates queries to all the directly connected monitor of it about the destination node. This process continues upwards. If any of them has it, then they may follow the reverse path to reply (R) the query made by the monitor. If multiple paths are found then the monitor takes the path from which the reply (R) came first. It
Fig. 2. Source is not directly connected to destination but monitor does
b) Now, if the monitor that is requested by a source node to send data on behalf that node, finds that the destination is not directly connected to it then searches the cache to find that route and once found data is send by that route.
264
III. A NALYTICAL M ODEL D
M
M
We assume that: 1) The duration of the packet arrival is an exponentially distributed with mean λ1 2) The time between location changes for each node is exponentially distributed with mean µ1 The probability [9] of a route broken is, µ PB = (1) (µ + λ)
M M M M
M S
M
Fig. 5.
Lemma 1: Let M be a monitor and Ni be any neighbor of the monitor then λ(M ) ≥ λ(Ni ) Proof: We have proposed that monitor is a node which has at least a predefined minimum number of neighbors and whose one hop connectivity is greater or same of all its neighbors. So the lemma holds. Lemma 2: If monitor M does not get any direct connection with the destination, D, then it must look for the route in its cache and if such a route exists, it delivers data towards that way- this is true if λ(M ) ≥ λ(S) and not necessarily λ(M ) > λ(D). Proof: The statement is true if and only if M is the monitor of S. But eventually even the node D can be itself a monitor. So the lemma holds. Lemma 3: Not necessarily but every node has a possibility of being a monitor, depending on the network state. Proof: Let, a node m is connected with node l and node n. If λ(n) > λ(m) and λ(l) > λ(n) then l is the monitor of m . Corollary 1: If λ(n) > λ(m) and λ(l) > λ(m) and if λ(m) = x, which is greater than the predefined number of first hop connectivity for being a monitor then m, n and l all are monitors to each other.
Dynamically route broken and route discovery
keeps the path in its cache. Here as the source request the monitor and the monitor finds that there is no such route in its cache, it then passes the packets to all neighboring monitors and via one of it finds the desired route (see Fig. 6).
D
M
M
M
R
M M R
M
R
M R
S
M
M Fig. 6.
Route query made by monitor discovery
e) If the monitor is querying for destination by asking all the neighboring monitors (M ). But the monitor Mk finds that it is in certain range from the source that exceeds the certain limit so that monitor Mk sends a ”destination unreachable” (U ) message to the source (see Fig. 7).
D
M
U
M M U
M
M M
U
U
IV. P ERFORMANCE S IMULATIONS To understand and evaluate the effectiveness of our proposed AMBR protocol, we have used a detailed simulation model of the NS-2 simulator [10], [11]. We keep the dimension of our simulation is 1300m×1300m. Traffic models were generated for 50 nodes and 100 nodes with CBR traffic sources, at a rate of 5kbps. The size of the packet is kept constant at 512 bytes. The pause time is varied anywhere between 50 sec to 350 sec with a interval of 50 sec, which is another measure of mobility. We compare the routing overhead as well as byte overhead against DSR-LRR, DSR, DSDV [7] and IZR for different pause time. Fig. 8 compares the different routing overhead packet for different pause time. We see that in all cases the number of routing overhead packet decreases as the pause time which resolves the mobility but it is much greater than our proposed protocol. In Fig. 9 we also compare the routing control bytes overhead against DSR-LRR, DSR, DSDV. In both the cases we found that the AMBR protocols generates lower
M S
M Fig. 7.
Monitor forwards ”destination unreachable” message
265
AMBR DSR-LRR DSR DSDV
Routing Control Traffic (Packets x 1000)
140 120
to the communication channel. So in brief, our proposed protocol is energy efficient.
100
V. C ONCLUSIONS In all reactive protocols, path determination and establishment depend upon extensive flooding that influences their effectiveness. In this paper, we proposed the adaptive monitor based routing technique which route a packet with the help of monitor and which repairs a route on the fly as soon as it is broken and eliminates the need for networkwide flooding. Using ns-2 simulator, we compared our routing scheme to DSR-LRR, DSR, DSDV and IZRP and observed that our technique resulted in a enormous reduction in packet and byte overhead. The trends in simulated overhead, together with the analytical model strongly indicate that AMBR is the only feasible ad hoc routing protocol. The proposed protocol improves other network characteristics as well. Possible future directions consists of introducing a technique from which a optimal neighbors number can be found for the selection of monitor for different network situations and to introduce a technique that will turn the partially periodic packets ”Monitor Alive Request” in to fully event driven packets.
80 60 40 20 0 0
50 100 150 200 250 300 350 400 Pause Time (Seconds)
Fig. 8.
Packet overhead versus pause time (50 Nodes)
routing overhead than the DSR-LRR [9], DSR, DSDV protocols.
Routing Control Traffic (Mega Bytes)
70
AMBR DSR-LRR DSDV DSR
60 50 40 30 20 10
R EFERENCES
0 0
50
100 150 200 250 300 350 400
[1] M.S. Corson and A. Ephremides, ”A Distributed Routing Algorithm for Mobile Wireless Networks,” ACM J. Wireless Networks, vol. 1, no. 1, pp. 61- 81, Feb.1995. [2] S. Murthy and J.J. Garcia-Luna-Aceves, ”An Efficient Routing Protocol for Wireless Networks,” ACM Mobile Networks and Applications J., vol. 1, no. 2, pp.183-197, Oct.1996. [3] T. Hara, ”Effective Replica Allocation in Ad Hoc Networks for Improving Data Accessibility,” Proc. IEEE INFOCOM, vol. 3, pp.1568-1576, 2001. [4] C. Perkins and E. Royer, ”Ad Hoc On-Demand Distance Vector Routing,” Proc. 2nd IEEE Workshop on Mobile Computing Systems and Applications, vol. 3, pp. 90-100, Feb.1999. [5] Prince samar , Marc R. Pearlman & Zygmunt J. Haas., ”Independent Zone Routing: An Adaptive Hybrid Routing Framework for Ad Hoc Wireless Networks,” IEEE/ACM Trans. on Networking, vol.12, no. 4, pp.595-608, Aug.2004. [6] M. Jiang et al., ”Cluster Based Routing Protocol (CBRP) Functional Specification (Internet-Draft), http://www.ietf.org/internetdrafts/draft-ietf-manet-cbrp-spec-00.txt, Aug.1998. [7] C. E. Perkins and P. Bhagwat, ”Highly Dynamic DestinationSequenced Distance-Vector Routing (DSDV) for Mobile Computers,” ACM SIGCOMM Computer Communication Review, vol. 24, no. 4, pp- 234-44, Oct.1994. [8] G. Pei, M. Gerla, and Z. Hong, ”LANMAR: Landmark routing for larfe scale wireless ad hoc networks with group mobility,” in Proc. IEEE/ACM MobiHOC, pp. 11-18, Aug.2000. [9] R. Duggirala, R. Gupta, Quing-An Zeng, and D.P. Agrawal, ”Performance Enhancement of Ad Hoc Networks with Localized Route Repair,” IEEE Trans. on Computers, vol. 52, no. 7, pp. 854- 861, July.2003. [10] The network simulator-2: http://www.isi.edu/nsnam/ns. [11] Paul Meeneghan and Declan Delaney, ”An Introduction to NS, Nam and OTcl scripting,” http://www.cs.nuim.ie/research/reports/2004/nuim-cs-tr-200405.pdf. [12] B. Chen and R. Morris, ”L+: Scalable landmark routing and address lookup for multi-hop wireless networks,” Technical Report MITLCS-TR-837, Laboratory for Computer Science, Massechusetts Institute for Technology, March 2002.
Pause Time (Seconds)
Fig. 9.
Byte overhead versus pause time (50 Nodes)
Routing Control Traffic (Packets X 1000)
We also compare the routing control overhead (see Fig. 10) of AMBR with IZRP [9] in the case of 100 nodes for different pause time. The routing overhead for IZR is same for different pause time and increase in pause time 250 and decreases after then but the routing overhead for AMBR is same for different pause time and it is much less than the IZR. 50
AMBR
40
IZR
30 20 10 0 0
Fig. 10.
50
100 150 200 250 300 350 Pause Time (Seconds)
Packet overhead versus pause time (100 Nodes)
From the graph we can easily realize that our protocol drastically cut down a large amount of routing control overhead. Smaller control traffic translates to lower power consumption, less congestion, smaller delays, reduced memory and processing requirements and faster access
266
