Experiments with Broadcast Routing Algorithms for Energy- Constrained Mobile Adhoc Networks. (Due in class on 7 March 2002)

Transcription

1 EE Project Description Winter Experiments with Broadcast Routing Algorithms for Energy- Constrained Mobile Adhoc Networks (Due in class on March ) Abstract In this project, you will experiment with the problem of energy efficient multicast routing in wireless Mobile Adhoc NETwork (MANET). Routing in these networks is a difficult problem because every node operates on limited battery resources and multihop routing paths are used over a constantly changing network environment due to node mobility. The network lifetime is defined as the duration of time until first node failure due to battery energy exhaustion. It is shown in [] that the network lifetime for a multicast session can be significantly extended by additionally considering residual battery energy as a parameter in cost metric function for constructing a power efficient routing tree. The goal of this course project is to () implement and compare shortest path spanning tree (SPT), minimum spanning tree (MST), and broadcast incremental power (BIP) [] algorithms over MANET, () improve the lifetime of networks. You are required to submit a report on your work not to exceed pages (excluding source code). You are advised to start early. Background Wireless Broadcast Advantage Compared to wired networks, wireless networks have broadcast advantage which is illustrated in Fig. S: Sender d M M, : Receivers M A: Receiver at the boundary Receivers inside the circle d q S d M S: Sender Figure : (a) Geometric construct of a sender S and receivers M and M (b) Wireless broadcast advantage Fig (a) shows a single sender S with receivers M and M at distances d and d, respectively, from the sender. We assume that d > d and the received power at a node varies as d i α (i =, ) where α is the path loss (attenuation) factor satisfying ( α ). Hence, the transmission power required to reach a node at a distance d i is proportional to d i α assuming the proportionality constant is. The second shows the broadcast nature of the wireless medium for omni-directional antenna in which a unit of message sent to receiver A at the boundary of the circle reaches every node within the circle for free. In order to transmit an identical message to nodes M and M, S can use two unicast transmissions with individual power d α and d α. This leads to the total expenditure of (d α + d α ) by S. However, the energy expenditure can be reduced by taking advantage of the fact that the wireless medium is naturally broadcast. Under this assumption, the sender has to choose between the following two strategies: (a) if d α > d α + d α, transmit to M and let M transmit to M, (b) otherwise, transmit to M directly (M will automatically receive it due to wireless broadcast advantage since d > d ). Hence, joint consideration of the effect of transmission and routing leads to savings in battery energy.

2 EE Project Description Winter BIP Algorithm For wireless mobile adhoc network, a power efficient algorithm called broadcast incremental power (BIP) algorithm [] is recently proposed which utilizes the wireless broadcast advantage properties. Input: given an undirected weighted graph G(N, A), where N: set of nodes, A: set of edges Initialization: set T := {S} where S is the source node of multicast session. Set P(i) := for all i N where P(i) is the transmission power of node i. Procedure: while T N do find an edge (i, j) T (N T) such that incremental power P ij = d ij α P(i) is minimum. add node j to T, i.e., T := T {j}. set P(i) := P(i) + P ij. Network Lifetime Extension Network lifetime is defined as the duration of time until the first node in a network fails due to the battery exhaustion. In case all the nodes have identical initial energy level, the node that spends the battery power at the highest rate will exhaust its battery first. If we want to extend the lifetime of the network, it is critical to incorporate the residual battery energy into route selection criteria. We note that, although the BIP algorithm produces a power efficient multicast routing tree for a single transmission of a packet (which is efficient for a short term period), it does not deal with maximization of the lifetime (which is a long term period) of a network. In order to incorporate the residual battery energy into the cost function, it is proposed in [] to multiply additional weighting factor to the incremental power P ij before constructing the total weighted cost function. The weighing function denoted by W i for node i is a time dependent function. The corresponding optimal tree is given by T arg min W P () WBIP i ij T G ( N, A) ( i, j) T n i Etotal ( Etotal k W E ) () ik and E total is the initial battery energy of node i, and E ik represents the amount of energy consumed at node i during the k-th update interval ( t). Therefore, the denominator of W i represents the remaining battery energy of node i at time t = n t. Notice here that the weighting factor W i is initially set to unity (i.e., BIP) and as time progresses and more energy is consumed, W i is monotonically increasing (i.e., W i ). The cost metric C ij W i P ij (WBIP) includes both node-based cost and link-based cost. The battery energy which is a characteristic of a node is represented in W i. The more a node has remaining energy, the less W i is and, therefore, there is a greater chance for this node with large battery capacity to be included in the route. The reason for W i being called node-based cost is that this value is equally weighted to all links to which this node is incident and it will be avoided if this weighting factor is large. On the other hand, P ij is a linkbased cost because different values are assigned for each link (i, j). Fig. presents a grid with α = for nodes using WBIP algorithm. The WBIP based routing tree solution at different time instances are shown in Fig. Every node is represented with a point and the solid lines correspond to the established links in the spanning tree. The circles with dashed-line represent the transmission ranges of the nodes located at the center of each circle. The remaining battery level is represented with a shaded rectangle in Fig.. The change of the route paths for different time instances is visually clear if considering the lower half of the network. It can be observed that the battery depletion is evenly distributed among the nodes by the choice of this metric, which justifies the metric () for lifetime extension.

3 EE Project Description Winter Broadcast routing tree based on BIP algorithm(total Power =.) at sec Broadcast routing tree based on BIP algorithm(total Power =.) at sec Broadcast routing tree based on BIP algorithm(total Power =.) at sec Broadcast routing tree based on BIP algorithm(total Power =.) at sec Figure : Routing trees at different time instances Implementation Tasks Table gives the geographic locations (x i, y i ) of hosts (nodes) for node ID i= participating in the network in x grid. Assume that there is no limit on the maximum transmission power P max. And the path loss α is set to be in this project. Table : Geographic locations (x i, y i ) of hosts (nodes) for node ID i= in x grid Node ID x i y i (source)

4 EE Project Description Winter Procedure ( Points) Implement shortest path spanning tree (SPT) algorithm (Dijkstra or Bellmanford), generate a final tree in graphic output. A sample graphic output for a final tree is shown as follows. Calculate the total power required to maintain the tree. BIP algorithm, Total Power =. Procedure ( Points) Implement minimum weight spanning tree (MST) algorithm (Prim or Kruskal), generate a final tree in graphic output. Calculate the total power required to maintain the tree. Procedure ( points) Implement broadcast incremental power (BIP) algorithm, generate a final tree in graphic output. Calculate the total power to maintain the tree. Procedure ( points) Given the initial battery energy level of each node is units (except the source node is K units) and constant bit rate traffic is adopted with rate = packet/sec. Each packet is transmitted with the corresponding transmission power determined by the algorithm. Let the update interval = sec for simplicity. Select one of three algorithms (SPT, MST, or BIP) and improve it in terms of network lifetime extension. You can use () as cost metric or propose any metric to extend the network lifetime. Fill out the table as follows. Time (sec) Total Power Outputs the initial routing tree and the final routing tree when the first node in the network fails, and compare the network lifetime of the original algorithm and the modified one using new cost metric. Procedure ( points) Compare the advantage and disadvantage of each algorithm when they are used in wireless ad hoc network environment. Discuss the shortcoming of these algorithms and how you conclude possible new directions to have optimal heuristic algorithms. Extract Credit Implement postsweeping algorithm. Programming Language You can use any programming language of your choice. However, MATLAB is strongly recommended not only because you can concentrate more on the concept not on implementation details but also you need to submit the final results with graphic output. When C/C++/Java/Fortran, etc is used, you need to submit the source code (including project files and auxiliary files) and binary executable files together. Any standard libraries (e.g. standard C/C++/Java library, STL, etc.) can be used but otherwise you should code from the ground up including data structures. Each source code should be well documented with rich comments. What and When to submit?

5 EE Project Description Winter () Report with all your simulation results and discussion. (Hard Copy, due in class on March ) A concise report of less than pages should be submitted. () Code (plus executable file if not in Matlab) (Hard copy attached to project report, soft copy sent to TA myli@ee.washington.edu by : am March ) Reference [] InTae Kang and R. Poovendran, On the lifetime extension of energy-efficient multihop broadcast networks, WCCI [] Wieselthier, J.E., Nguyen, G.D. and Ephremides, A. (), On the construction of energy-efficient broadcast and multicast trees in wireless networks, Proc. IEEE INFOCOM, pp. --.