G-REMiT: An Algorithm for Building Energy Efficient Multicast Trees in Wireless Ad Hoc Networks Bin Wang and Sandeep K. S. Gupta NCA’03 speaker ： Chi-Chih Wu

Outline Introduction System Module and Assumption Multicast Energy Cost Estimation G-REMiT Protocol Details Performance Evaluation Conclusions

Introduction Wireless ad hoc network (WANET) – –myriad wirelessly connected static devices – –no infrastructure Energy efficiency is an important design consideration in WANET – –serious resource (battery) constraints

Introduction Multicast is an important communication primitive – –Group communication – –Distributed cache maintenance Multicast tree is an efficient approach to overcome resource wastage in multicasting

System Module and Assumption Assumptions – –Each node in the WANET has only local view of the network – –Each node knows the distance between itself and its neighbor nodes using distance estimation schemes – –We assume that every node can know the number of nodes in the multicast group – –Average of multicast message generation rate at a given node is stable over a period time.

System Module and Assumption p i,j = E elec + K( r i,j ) α –p i,j ： minimum power needed for link between nodes i and j –E elec ： –E elec ： distant-independent constant that accounts for realworld overheads of electronics and digital processing – –K ： constant dependent upon the properties of the antenna –r i,j –r i,j ： Euclidean distance between i and j – –α ： constant which is dependent on the propagation losses in the medium L. M. Feeney and M. Nilsson. ， ”Investigating the energy consumption of a wireless network interface in an ad hoc networking environment”

System Module and Assumption MST (minimum weight spanning tree) multicast group members forwarding nodes R. Gallager et al. ， ”A distributed algorithm for minimum weight spanning trees”

Multicast Energy Cost Estimation from the energy cost at every tree node is not only decided by the distance between connected tree neighbors but also decided by where the message is coming from Max(p9,1 ， p9,3 ， p9,4 ， p9,10) = p9,3 Max(p9,1 ， p9,2 ， p9,3 ， p9,4) = p9,2

Multicast Energy Cost Estimation p i,j = E elec + K( r i,j ) α λi ： Message generation rate

Multicast Energy Cost Estimation p i,j = E elec + K( r i,j ) α

Multicast Energy Cost Estimation

G-REMiT Protocol Details Criterion for switching –E elec = 0 –α = 2 –Change 2 9,6 –Gain (g 2 9,6 ) –R 2 = r 9,2 2 = R 6 = r 6, * r 6,8 2 = R 9 = r 9, * r 9,2 2 = –R 2 ’ = r 2,6 2 = R 6 ’ = r 6, * r 6,8 2 = R 9 ’ = r 9, * r 9,3 2 = –g 2 9,6 = (R 2 +R 6 +R 9 ) – (R 2 ’+R 6 ’+R 9 ’) = g 2 9,10 = (R 2 +R 9 +R 10 ) – (R 2 ’+R 9 ’+R 10 ’) = g 2 9,8 = (R 2 +R 9 +R 8 ) – (R 2 ’+R 9 ’+R 8 ’) = 44.82

G-REMiT Protocol Details Algorithm description –Distributed –Divided into two phases First phase –Run Gallager et al. ， ” –Run Gallager et al. ， ” A distributed algorithm for minimum weight spanning trees ” ， to build a MST tree – –Nodes have all local information

G-REMiT Protocol Details

Second phase –Depth-First Search (DFS) –Token Example 1 ： i x j Join_REQ Join REP

G-REMiT Protocol Details Example 2 ：

Performance Evaluation 10, 20, 40, 60, 80, 100, 200 nodes The graph is connected Network bandwidth is 50 kb/s Message generation rate is randomly generated as 0, 1, 2, …, 50 kb/s α = 2

Performance Evaluation TPC (total power consumed) MIP 、 MLiMST – –J. Wieselthier et al. ， ” On the construction of energy- efficient broadcast and multicast tree in wireless networks ” ， INFOCOM 2000

Performance Evaluation

Conclusions G-REMit algorithm to construct an energy- efficient multicast tree G-REMit algorithm performs much better than most existing energy efficient multicast algorithm In feature ， we plain to extends G-REMit algorithm to mobile ad hoc networks and study its performance