1. 1 A Multi-Rate Routing Protocol with Connection Entropy for MANETs Cao Trong Hieu, Young Cheol Bang, Jin Ho Kim, Young An Kim, and Choong Seon Hong Presenter: Cao Trong Hieu Department of Computer Engineering, Kyung Hee University, Korea hieuct@networking.khu.ac.kr Jan, 15, 2010

2. 2 Outline  Motivation and Main Contributions  Link Weight Definition for Muti-rate MANETs  Modeling Connection Uncertainty of a Route  The Multi-rate Routing Metric  Performance Analysis  Conclusions

3. 3 Motivation and Main Contributions  For wireless communication: either relative distance or the path loss will effect the data rate between two communicating nodes.  For mobile multi-rate ad hoc networks: the connection certainty of a link also changes because a node may move closer or far away out of communication range of other nodes.  Need to measure the frequent change of distance  relate to connection entropy and stability of a route.  Contributions:  New approach to model node mobility based on link rate adjustment.  New metric consider both link weight and route stability that be defined in relation with connection entropy.

4. 4 Data Rate and Rx Sensitivity  Nowadays, physical layer enhancements support multiple data rates.  For example, the IEEE 802.11g standard with OFDM technology support 8 data rates according to the selected Modulation and Coding Scheme (MCS) as showed in the Table 1.

5. 5 Link Weight Definition for Muti-rate MANETs  The received signal strength at receiver R with distance d far away from the transmitter T is calculated as:  Let Pr = Rx sensitivity, we have:  Definition: Link Weight

6. 6 Modeling Connection Uncertainty of a Route  Consider network with n nodes, l is the maximum number of links of possible routes from the source node s to the destination node d (l≤n-1)  Let be connection probability that take the value [0,1] (represents the probability that node j-th stays at the i-th position in the route)  The probability distribution is complete for a legitimate route, hence:  is the (l x n) matrix of connection probabilities. In which

7. 7 Modeling Connection Uncertainty of a Route  For a route, the connection entropy is defined as:  The connection entropy reflects the stability condition of MANET where nodes are moving: If H(P) = H min = 0, the route is least uncertainty. If H(P) = H max = ln n, the route is most uncertainty.  Unstable route with high connection uncertainty should be avoided  need to minimize H(P)

8. 8 Associating Link Weight with Connection Uncertainty  The link weight of a node defined previous exists with probability of connection  For a route, the weight associated with the i-th position is the sum of the link weight between the node that occupies the i-th position and the nodes occupying the previous and next positions in the route.  The average link weight for each position in the route:

9. 9 The Multi-rate Routing Metric  Depend on the instant position of a node with the nodes occupying the previous and next positions in the route, the corresponding link weight can be determined.  When nodes moving  relative distance is changed  communication rate and connection entropy are changed.  The multi-rate routing metric considers both link weight and connection entropy:  Minimize M(,P), subject to minimize H(P) and maximize

10. 10 Performance Analysis  100 nodes randomly distributed in 250mx250m area  802.11g standard was used with 8 supported rates (6, 9, 12, 18, 24, 36, 48, 54)Mbps  Random way point mobility model with maximum speed 10m/s  The path loss exponent  (range 2 to 6) is set to 3  UDP traffic with packet size 1000 bytes  Weighted parameter  = (0, 0.5, 1)

11. 11 Performance Analysis  Comparison of route discovery time, percentage of link breakage with AODV

12. 12 Performance Analysis  End-to-end throughput

13. 13 Conclusions  For the MANET, routing metric must select not only high throughput but also stable route for delivering data.  By modeling mobility aspect as the variation of the level of maximum data rate that connecting nodes can use when they move close to or far away from other nodes, we relate mobility problem with link weight and route stability.  New routing metric considers both high throughput and route stability.  For the future, consider multi-path routing for multi-rate network. More analysis and comparison with other routing metrics need to be done.

14. 14 THANK YOU!