Você está na página 1de 4

A Routing Algorithm for Supporting Soft-Qos in Mobile Ad hoc Networks

Xiaonan Feng
Institute of Electrical and Electronic Engineering North China Electric Power University Beijing 102206, China fengxa@ncepu.edu.cn
AbstractRecent work in mobile ad hoc network routing protocol development has demonstrated how global route discovery can be performed more efficiently by leveraging the known topology of each nodes local surrounding area. In this paper, we proposed a hybrid routing scheme based on a dynamic clustering algorithm. The adaptive clustering algorithm deals with mobility, transmit power conservation and improving system throughput. Routing algorithm proposed in this paper is decoupled from the clustering algorithm, it dynamically balance the tradeoff between proactive and reactive routing, while also dynamically tradeoff route optimality for routing overhead in a novel way. Moreover, this paper describes how to gather three metrics, which are used to support Qos aware route computation. Keywords-routing algorithm; Soft-Qos;mobile ad hoc network

in order to adjust that logical topology and make clustering decisions. Routing protocol proposed in [7] mainly can reduce average end-to-end delay and loss rate, does not describe how to update the routing status. In order to limit routing overhead, routing policy for ad hoc wireless network has progressed from proactive, reactive to hybrid routing. Proactive routing protocols attempts to maintain shortest-path routes by using periodic updates to track changes in the network topology. The main drawback of this technique is that there are routing updates even when there are not needed, such as when nodes have not moved or are not involved in communication. Reactive routing protocols provides the basic motivation for the on-demand routing protocol, which create and maintain routes only when required, thus saving on redundant routing overhead from periodic updates. Pure adopts proactive or reactive routing protocols can not completely solve the problems. In this paper, we proposed a hybrid routing protocol utilizes a dynamic clustering algorithm that is adaptive with respect to node mobility. The adaptive cluster organization supports a hybrid routing approach dynamically balance the tradeoff between proactive and reactive routing, while also dynamically tradeoff route optimality for routing overhead in a novel way. II. NETWORK MODEL AND DEFINITION

I.

INTRODUCTION

The routing problem in ad hoc networks is complicated since any, or all, of the hosts involved may move at any time. Numerous challenges must be overcome to realize the practical benefits of ad hoc networking [1], for the network is highly dynamic and transmissions are susceptible to fading, interference, or collision from stations. A central challenge in the design of ad hoc networks is the development of dynamic routing protocols that can efficiently find routes between two communicating nodes. The routing protocol must be able to keep up with the high degree of node mobility that often changes the network topology drastically and unpredictably, but routing overhead is the critical factor impacting performance of routing protocol. In [2], a clustering management mechanism have developed to adjust the mobile terminals transmission power to smooth the drastic topology change based on current mobility scenarios. Nasipuri et al in [3] use preemptive search by studying the statistical distribution of the link and route lifetimes to balance the optimality of the routes as well as the overhead incurred from transmit routing packet. In [4], a QOS-based robust multipath routing protocol is proposed to accomplishes increased packet delivery ratio with reduced latency. In [5], Michail and Ephremides aimed to minimize energy expenditure in addition to minimizing the overall blocking probability. Hybrid adaptive routing strategy proposed in [6] is using a path stability to cluster network, the clustering algorithm presents a logical topology to the routing algorithm, and it accepts feedback from the routing algorithm

1) Network model In a wireless network, each node has a transceiver for communication. The neighbors of a node, which a node can directly communicate with, is not fixed but depends on the power used by its radio transmitter. When the power of the radio transmitter is increased, a node can directly communicate with a larger set of nodes. The propagation function is represented as R: LLZ, where L is a set of location coordinates in the space. Function (li , l j ) gives the loss in dB due to propagation at location

l j L , when a packet is originated from location l i L .


The successful reception of a transmitted signal depends,

tP (li , l j ) S

(1)

978-1-4244-6252-0/11/$26.00 2011 IEEE

Along with the propagation function R, on the transmit power tP of sender and the receiving sensitivity S of receiver. The receiving sensitivity is the threshold signal strength need for reception and is assumed to be a previously known constant, same for all nodes. In particular, for successful reception, we assume that R, is a monotonically increasing function of the geographical distance

Definition 2 Residual power This metric is important in wireless network because most portables are powered by batteries with limited weight and life. Furthermore, the remaining power is essential to maintain a stable network and achieve good communication quality. The remain power will be get by the formula:

d (li , l j ) between l i and

l j . This is generally true for free space propagation. We can


then combine S and R, into one function as fellows,

resP = aP

average(tw)

(5)

(d ) = (d (li ,l j ))+ S

(2)

Clearly, tP must be at least (d ) for successful reception. So given an Ad hoc network M=(N, L)(N is the nodes of network and L is the link between the nodes), a transmit power function p, and a least-power function , we can represent the induced graph as G=(V, E), where V is a set of vertices corresponding to nodes in N, and E is a set of undirected edges so that (u, v)E if and only if p(u) (d(u,v)), and p(v) (d(u,v)). Then, we can think of the links as being symmetric and the resulting graph as undirected. Furthermore, we assume that there exists a medium access scheduler so that each node can transmit at a certain bit rate without interference. 2) Routing parameters Before given routing metrics, we give some abbreviations in table I for simplifying next descriptions
TABLE I. tP(w) rP(w) AP(w) Tw(w) ABBREVIATIONS

When some nodes leave far from each other, then the nodes will increase their power of communicating. On the other hand, if the nodes get closer, then the power of communicating will decrease and thus save the valuable power resource. Definition 3. Available Bandwidth In adaptive clustering for mobile wireless networks, the available bandwidth is limited. The first data packet of the flow makes reservations along the path. Once the packet is received, a transmission window is reserved for the subsequence packets in the connection. The window is released until several idle cycles occur. If the link isnt broken due to mobility, the subsequence packets will be received successfully. The bandwidth is given by: B=[cycle time/ frame time] And the available Bandwidth (aB) can be as:

aBi = mBi jN uB j
i

(6)

Transmit Power Receive Power sensitivity Residual Available Power Estimation of Total power consumption based on the last periods

Where mB denotes Maximum Bandwidth of node capability and uB denotes Used Bandwidth (uB) for active connection. III. HYBRID QOS ROUTING ALGORITHM(HQRA)

For improving Qos routing guarantee intensity, we will take the following three parameters as main reference. Definition 1. Aggregate mobility metrics Presented in [8], the aggregate mobility metriccs is used to meet the biggest requirements of QoS stricter application. The relative mobility metric is defined as:
new rPi j

mRi j = 10 log10 (

tPjnew i

old rPi j

(3)

tPjoldi
Then calculate the aggregate local mobility valued:

Here we propose a hybrid routing algorithm. It is the combination of proactive policy and reactive policy. It divides the nodes into several clusters through suitable size clustering algorithm(SSCA) in [2]. Based on shis algorithm intra-cluster routing use a proactive policy, whereas the inter-cluster routing is reactive. In networks with low rates of mobility, clustering provides an infrastructure, which is more proactive. This enables more optimal routing by increasing the distribution of topology information when the rate of change low. When mobility rates become very high. Cluster size will diminish and reactive routing will dominate. Decoupling the routing algorithm specification from the clustering algorithm, thus, it is flexible enough to support evolving ad hoc network routing strategies in both intra and inter-cluster domains. The hybrid policy accompanies a better balance between power saving and quick routing. When a node needs to discover a route to the destination, except considering the least time-consuming route, stability also should be take into consideration. Thus those nodes which have enough power and bandwidth remain should be give some priority. The routing update includes two procedures, intra-cluster routing update and inter-cluster routing update. Intra-cluster

mRi = var mRi j

jN i

(4)

A node that has the lowest M among its neighbors denotes it has more stable connection to its neighbor, also means it is with easier to find alternate link for abrupt disconnection than other nodes.

routing update cycle is shorter than that of inter-cluster routing cycle. Different from [9] adapting to the mobility by adjusting cluster size, our scheme decouples clustering and routing, routing algorithm adjusts the inter-cluster update scope to adapt the mobility. Fig. 1 briefly describes the relationship between clustering algorithm SSCA and routing algorithms HQRA.
Proactive Routing

HQRA
Topology

Reactive Routing

Fig. 2 illustrates the process of above procedure. Node A initiates an inter-cluster route update package and sends to node B transferred by node 1 (similar to node C and D), if node B has received inter-cluster update package of the cluster delegated by node A in the successive inter-cluster update periods, node B forward this update package to node E and F, so node E and other nodes which is in the same cluster with E know the topology and link status of the cluster delegated by A. On this time, local routing table of node E at least includes the status of three clusters, which delegated by E, B, F respectively. The inter-cluster routing information delivers to inner nodes in the intra-cluster update procedure. But only intracluster route status is packed into the inter-cluster update package. For example, in figure 2, when node A initiates intercluster update package it does not include the route status of node B though its local route table containing available route to B. IV. PERFORMANCE EVALUATION

SSCA
Figure 1. Relationship between Clustering and Routing

The intra-cluster routing update is implemented by clusterhead. The clusterhead sends intra-cluster route status packet to its cluster members periodically, and the members will update their routing table after they receive the packet. The inter-cluster routing update can employ any proposed proactive routing schemes. Any clusterheads designate an internal node as inter-cluster routing updater by some criteria, such as maximal residual power. These updaters execute intercluster routing update procedure as follow: 1) Inter-cluster Updater initiates intra-cluster route status package and send it to its direct neighbors cluster periodically. 2) Any inter-cluster Updater receiving a inter-cluster update package performs below actions: a) Integrates the routing status information of this package to its local routing table and records the updating path of the source cluster. b) Refreshes timers of routing table items according to the new arrival inter-cluster update package. c) Checks the travel path of this package with that of last update. If successive update packages initiating from the same cluster have traveled on the same cluster path, the updater forwards new update package to its direct neighbors except of the coming cluster, Or else no forwarding is performed. 3) The timeout route items are removed from local routing table when its timer event arrives.
B 3 E 1 A 2 5 D F 7 G C 6 H

Topology

A. Simulation Environmentt We implemented our routing scheme within the Glomosim library [10]. The Glomosim library is a scalable simulation environment for wireless network system using the parallel discrete-event simulation language called PARSEC [11]. Our simulation models a network within 2000*2000 meter square and the nodes in the network is placed uniformly. Radio propagation range for each node is 150 meters and channel capacity is 2 Mbits/sec. In most of experiments unless specified, the network consists of 100 nodes and the average movement speed varies from 5m/s to 45m/s. Each simulation executed for 600 seconds of simulation time. We run each scenario three times and the data collected are averaged over those runs. B. Simulation Results We compare the routing overhead between Fisheye and HQRA from two aspects, number of nodes and node mobility speed. The following figures are the simulation results showing the impact of routing overhead from number of nodes and the node mobility speed.
HQRA Fisheye

1000 900 800 700 600 500 400 300 200 100 0 50 60 70 80 90 100 110 120 130 140 150
Number of Nodes

Figure 2. Example of inter-cluster routing update Figure 3. Overhead comparison by number

Average Control Overhead

HQRA 700 Average Control Overhead 600 500 400 300 200 100 0 5 10 15

Fisheye

clustering schemes. By adjusting the inter-cluster update scope according to network dynamic degree to get the goal of minimizing routing overhead. In addition, it calculates Qos aware status parameters and updates to foreign clusters for support local Qos route computation. The future work is needed to assess the performance of this algorithm combined with routing selection scheme. REFERENCES
[1] Chakrabarti, S., Mishra, A., Qos Issues in Ad hoc Wireless Network, IEEE Communications Magazine, Feb. 2001, pp. 142148. [2] Xin Jin, Hongbo Wang, Yaoxue Zhang, A cooperative two-tier framework for efficient routing in MANET, IEEE Proceedings of International Conference on Computer Networks and Mobile Computing(ICCNM), Oct. 2003, pp. 465469. [3] Nasipuri, A., Burleson, R., Hughes, B., Roberts, J., Performance of a Hybrid Routing Protocol for Mobile Ad hoc Networks, IEEE Proceedings of Tenth International Conference on Computer Communications and Networks, Oct. 2001, pp. 296302. [4] Venkatasubramanian, S.,Gopalan, N.P., A QoS-based robust multipath routing protocol for mobile ad hoc networks, IEEE First Asian Himalayas International Conference on Internet(AHICI), Nov. 2009, pp. 17. [5] Michail, A., Ephremides, A., Energy Efficient Routing for ConnectionOriented Traffic in Ad hoc Wireless Networks, The 11th IEEE International Symposium on Personal Indoor and Mobile Radio Communications (PIMRC), 2000, Vol. 2, pp. 762766. [6] McDonald, A.B., Znati, T., A dual-hybrid adaptive routing strategy for wireless ad-hoc networks, IEEE Wireless Communications and Networking Conference (WCNC), 2000. Vol. 3, pp. 11251130. [7] Yanmei Yang, Wendong Han, A QoS Routing Protocol for Mobile Ad Hoc Networked Control Systems. IEEE Second International Conference on Networks Security Wireless Communications and Trusted Computing (NSWCTC), Jun. 2010. pp. 8992. [8] Basu, P., Khan, N., Little, T.D.C., A Mobility Based Metric for Clustering in Mobile Ad Hoc Networks, IEEE International Conference on Distributed Computing Systems Workshop (ICDCS), Apr. 2001, pp. 413418. [9] Lin, C.R., Chung-Ching Liu, An on-demand QoS routing protocol for mobile ad hoc networks, IEEE International Conference on Networks (ICON), Sep. 2000, pp. 160164. [10] Gerla, M., Tsai, J.T.C., Multicluster, mobile, multimedia radio network, Wireless Networks, vol. 1, 1995, pp. 255265. [11] Malnar, M.Z., Neskovic, N.J., Comparison of ETX and HOP count metrics using Glomosim simulator, IEEE 9th International Conference on Telecommunication in Modern Satellite, Cable and Broadcasting Services(TELSIKS), Oct. 2009, pp. 8588.

20 25 30 Number of Speed

35

40

45

Figure 4. Overhead comparison by speed

Fig.3 reports the comparison of routing overhead between Fisheye and HQRA by the number of nodes. When the number of nodes gains, the routing overhead of the two algorithms both increased. But the routing overhead of HQRA is always lower than Fisheye. Because in Fisheye, every node makes itself the centre of a circle and sends routing update message to other nodes with the frequency corresponding to the scope radius, while in HQRA, the routing update process is based on cluster. So when the number of nodes increased only a part of new nodes participate in the routing update process. Therefore the routing overhead of Fisheye is increasing faster than that of HQRA. Especially when the number of node is more than 100, HQRA reduces more than 40% of routing overhead compared with Fisheye. Fig.4 reports the comparison of routing overhead between Fisheye and HQRA by the node mobility speed. When the node mobility speed is slow the routing overhead of HQRA is much lower than Fisheye. With the speed increasing, the routing overheads of HQRA and Fisheye are both increasing, but even in the worst case that the node mobility speed is very fast and the cluster can not be maintained, because the HQRA can dynamically reduce the scope of routing update, so the routing overhead of HQRA is still lower than that of Fisheye. V. CONCLUSION

In this paper, a hybrid QoS aware routing update algorithm has proposed to minimize the total routing overhead. It is based on an adaptive clustering algorithm but decouples from

Você também pode gostar