1. On the Construction of Energy- Efficient Broadcast Tree with Hitch-hiking in Wireless Networks Source: 2004 International Performance Computing and Communications Conference (IEEE InfoCom ) Author : My T. Thai / Yingshu Li / Ding-Zhu Du Repoter : Yen-Lin Chen Date : 2005/03/23

2. Outline  Introduction  Preliminaries Communication Model Hitch-hiking Model Network Model  Related Work  The Broadcast with Hitch-hiking Algorithm  Simulation Results  Conclusion

3. Introduction  In this paper, we study the problem of minimizing the total broadcast energy in wireless networks.  The broadcast problem in wireless networks is to decide a transmission power level for each node so that the source node can broadcast to all the other nodes.

4. Introduction (Cont.)  Our objective is to construct a minimum- power broadcast tree, rooted at the source node, including all the nodes.  Our key idea is to reduce the total energy consumption of the broadcast tree by taking the Hitch-hiking model and Wireless Multicast Advantage (WMA) into consideration.

5. Introduction (Cont.)  By successfully combining partial signals to obtain complete information, we can efficiently reduce the total energy consumed in broadcasting data.

6. Communication Model  Assume that any node in the network can be used as a relay node to forward data to other nodes. All nodes are equipped with omnidirectional antennas.  Assume that all nodes can adjust their power levels. Each node can choose its transmission power from 0 to some maximum value P max.

7. Communication Model (Cont.)  Attenuation model: d : the signal travelling distance α : an environmentally dependent real constant between 2 and 4 P i : power level, a node j can properly receive a signal from a node i. γ : represents the receiver's power threshold for Signal to Noise Ratio (SNR), often normalized to 1.  For node i in the network, the power required to successfully transmit data to node j is given by:

8. Hitch-hiking Model  Introducing two thresholds on the SNR: Threshold energy required for the successful decoding of the message Threshold energy required for the successful capture of a packet Usually  A packet received with a SNR γ is: Full reception if Partial reception if Failed reception if

9. Hitch-hiking Model (Cont.)  If a node receives the packet containing the same information n times from different neighbors with such as Assume that ° for simplicity. Node can successfully receive the packet.

10. Network Model  Let V denote the collection of wireless nodes  Let G = (V,E) denote the directed graph on V that contains all edges.  Every node has an associated transmission power level.  : node i transmits a packet, the amount of reception by node j is quantified by the coverage of node j.

11. Network Model (Cont.)  The coverage function :  The coverage provided by node i on node j is:

12. Related Work  Broadcast Incremental Power (BIP) Similar to Prim’s algorithm for forming minimum spanning tree Weights are dynamically updated at each step  The BWHH algorithm can be obtained from BHH with a change on the coverage function. The coverage provided by node i on node j is defined as:

13. Related Work (Cont.)  Wireless Multicast Advantage (WMA) That is a single transmission can be received by all the nodes that are within the transmission range, reduces the total energy of the broadcast tree. Nodes have omnidirectional antenna ‘i’ transmits at and reaches both j and k Energy expenditure j k i p ik p ij

14. The Broadcast with Hitch-hiking Algorithm S(8) U(5) T(5) Y W(2) X Q V(1) Z S(10) T(1.9) Y W X Q V Z U(2.4) (0.48) (0.24) (1) (0.55) (0.76) (0.62) (0.38) Start with Minimum Spanning Tree Improve upon the initial solution starting with source node  At each step, pick a fully covered node (Ex: node ‘ S ’ ) and decide its power level at which the gain is maximum  Calculate the coverage of all other uncovered nodes based on the final power level of node ‘ S ’ (Ex: node X,Y and Z)  Calculate the power level of forwarding nodes based on new coverage value of their child nodes (Ex : node U, T and V)

15. The Broadcast with Hitch-hiking Algorithm (Cont.)  Let us first introduce the following notations: c(ji): the coverage of node i provided by node j. c(ji) is defined in formula (1) c(i): the total coverage of node i provided by all the other nodes. c(i) = P h (i): the power level of node i at the iteration h N(i): the neighbors of node i P h : be the set of nodes that are not fully covered in the network at the iteration h. F h = V – P h :are fully covered in the network at the iteration h.

16. The Broadcast with Hitch-hiking Algorithm (Cont.)  Let us define a ratio r( i ):  p h ( i ) – p h- 1 ( i ) is the incremental power of node i  is the sum of the updated coverage of all nodes j after increasing the power level of node i where j is in the partially covered node set of the previous iteration, p h-1.

17. The Broadcast with Hitch-hiking Algorithm (Cont.) Figure 1(a) represents the network where the maximum power level of node S and A are 10 and 5 respectively.

18. The Broadcast with Hitch-hiking Algorithm (Cont.) Figure 1(b) shows the broadcast tree T at the h iteration. At this step, P h = {B,C} and F h = {S,A}.The ratio r of node S when increasing the power level of node S to fully cover node B is:

19. The Broadcast with Hitch-hiking Algorithm (Cont.)  In order to minimize the total power consumption,we not only want the incremental power of a node at each step to be small, but also want the sum of the coverage of all nodes to be large.

20. Simulation Results  We evaluate the performance of BHH by comparing it to another two algorithms, BIP and WMH.  In this simulation, we considered the following parameters: n: the number of nodes in a network. Thereby increasing the network density when the number of nodes increases. n is from 10 to 100. P max : The maximum power level of each node is randomly assigned on each simulation setup.

21. Simulation Results (Cont.)

22.  The total power of the broadcast tree constructed using BHH is almost 77% less than that of BIP, and 15% less than that of WMH.  The power of the broadcast tree constructed using BWHH is 49% less than that of BIP.  The total power of a broadcast tree decreases when the number of nodes increase.  Because as the network density increases, more nodes are available to work as relay nodes and the nodes are becoming closer.

23. Simulation Results (Cont.)

24.  We present the comparison of the improvement on saving energy over another three algorithms in percentage: BIP vs. BHH, BIP vs. BWHH, WMH vs. BHH.  As can be seen in this Figure, BHH is the one that can save the most energy, comparing to both BIP and WMH.  This analysis indicates that combining WMA and Hitch-hiking model does achieve a better result.

25. Conclusion  We proposed the BHH algorithm based on the Hitch-hiking model.  This algorithm takes advantages of WMA and of the Hitch-hiking concept.  It is our interest to further develop the distributed version of BHH in mobile environment.