160 likes | 265 Views
Load Balancing Routing Scheme in Mars Sensor Network. CS 215 Winter 2001 Term Project Prof : Mario Gerla Tutor: Xiaoyan Hong Student : Hanbiao Wang & Qingying Li. Objective. Balancing sensors’ energy consumption by diffusing data traffic into their closest neighbourhoods.
E N D
Load Balancing Routing Scheme in Mars Sensor Network CS 215 Winter 2001 Term Project Prof : Mario Gerla Tutor: Xiaoyan Hong Student : Hanbiao Wang & Qingying Li
Objective • Balancing sensors’ energy consumption by diffusing data traffic into their closest neighbourhoods. • Prolong network lifetime by alleviating the load unbalance problem.
Problem of Original Design for JPL Sensor network • In the original implementation, it forms a spinning tree for routing path • Node 35 Handle data traffic for cluster 1 • Node 74 Handle data traffic for cluster 2 • Conclude : Node 35 and 74 died out fast Cluster 1 74 35 Cluster 2
Proposed Algorithm : We propose an multi path routing scheme to diffuse data traffic of the sensor to its neighbourhood that is still in the shortest path back to the base station. • Routing table construction stage, the sensor network self-organized into a configuration of N levels. • a) The base station is the sink with level = 0 ( black in the figure) b) All the nodes that can directly hear from the base station are labelled as level = 1 ( blue in the figure ). All the nodes that could directly talk with level 1 nodes are labelled as level = 2 (green in the figure), but they couldn't directly talk with level 0 node, and so forth. • data transmission step, node i at level (N+1) will randomly chooses next hop j from its neighbours at level N with equal probability.
Design Issue 1 : Construct Routing Table When a node receives request packages from other nodes,it will check it against its own routing table: • HopNumber ( Incoming packet) > HopNumber( record ) discarded ; • HopNumber(Incoming packet) < HopNumber ( record ) delete old path and record new path; 3. HopNumber(Incoming packet) = HopNumber(record) insert this new path into its routing table
Example for Node i • Assuming Node i received a package from Node J at first. The node J indicates that its distance to the base is 3, hence, Node i will record its distance to the base is 4 ; • Then another package from K arrives and indicates its distance to the base is 2. Hence, Node i compare this with its record, it will delete the path via Node J and update its distance to the base as 3, also record the path via node K. • When the package from Node L arrives and indicates its distance to the base is 2. Node i will add this entry to its routing table that gave same shortest distance of 3 to the base, via Node L. • Node M has same situation as Node L.
Flooding request package Level 2 2 Level 1 2 1 2 2 3 Base 1 1 1 J 2 K 1 1 2 2 i L 3 2 2 Level 3 Level 4 M
Data Transmission from Node D To Base Station • Node D starts to send data package to the base station via Node i since it is the only path in its routing table. • Node i will randomly pick one of the entry ( choose the next hop from routing table entry, via K, or via L, or via M) with equal probability to determine which path to forward the data. • Assuming K is chosen by Node i and now node K will check its routing table, determine its next hop, either via Node X or Node Y. • Suppose X is chosen and its next hop is the base station and the package is now transmitted from Node D to the base station. • During this Data Forwarding, we used routing path 1
4 Transmitting Data from Node D to Base with 4different path: Path 1 -- Grey Path 2 -- Red Path 3 -- Green Path 4 -- Yellow 2 2 1 2 1 2 2 X Base 1 1 1 3 2 1 1 K 1 2 2 Y 2 i 3 3 L 2 2 D 3 M 4
Multi Path Nodes for Senor Network • Level 0 Black • Level 1 Blue • Level 2 Green • Level 3 pink • Level 4 white • Level 5 yellow • Level 6 Grey
Multi Path for Sensor network • Node has alternative path to send data to base station.
100 nodes with 2 Hour Simulation Result Level 3 Nodes Power consumption • Red Curve = nodes’ power consumption in original implementation in JPL sensor network • Blue Curve = nodes’ power consumption in multi path design. • Original power consumption Range :{110.29w,260.27w} • Our power consumption Range:{116.73w, 208.44w} • No packages lost in new implementation
100 Nodes with 2 Hour Simulation Result Level 4 Nodes Power consumption • Red Curve = nodes’ power consumption in original implementation in JPL sensor network • Blue Curve = nodes’ power consumption in multi path design. • Original power consumption Range :{92.31w,186.16w} • Our power consumption Range:{95.84w, 151.63w} • No packages lost in new implementation
200 nodes with 5 Hour Simulation Result Level 1 Nodes Power consumption( test when the nodes are more dense) • Red Curve= nodes’ power consumption in original implementation in JPL sensor network • Blue Curve = nodes’ power consumption in multi path design. • Original power consumption Range :{373w,1272.35w} • Our power consumption Range:{408.61w,1149.19 w} • No packages lost in new implementation
Conclusion and Future work Advantage : 1) Balancing Node Work Load 2) Prolong network life time 3)No performance disgrade Future Proposal : Data transmission : when Node I of level N+1 tries to pick next Hop J of level N, it should chooses according to : Probability (I picks J) E j / Pij E j = Energy Left in Node J Pij = Power needed for transmitting data from I to J
Reference • Xiaoyan Hong et al., The Mars Sensor Network: efficient, power aware communications, (Milcom 2001) • Chalermek Intanagonwiwat, Ramesh Govindan, and Deborah Estrin, Directed Diffusion: A Scalable and Robust Communication Paradigm for Sensor Networks, Proceedings of the Sixth Annual International Conference on Mobile Computing and Networks (MobiCOM 2000), August 2000, Boston, Massachusetts • Marc R. Pearlman et al., On the impact of Alternative Path Routing for Load Balancing in Mobile Ad Hoc Networks, MobiHoc 2000. • Ya Xu, John Heidernmann and Deborah Estrin, Geography-informed Energy Conservation for Ad Hoc Routing., MobiCOM 2001.