Topology Control in Wireless Sensor Networks

1. Topology Control in Wireless Sensor Networks

2. Three R&D Styles • Intuitive approach (e.g., directed diffusion) • Easy to understand, a lot of follow-up work • refreshing ideas make you happily surprised • Need a BIG and innovative idea • Easy to attack since everyone can understand the work quickly • Analytical approach (e.g. this homework) • Solid work, summarizing or generalizing previous pieces of work • Novelty may be limited, practical impact?? • System approach (e.g. the modular net layer paper) • Practical impact is HUGE • Insights learned from system building and user experiences • No new ideas that will blow your mind away • A collection of small ideas or tricks

3. Coverage and Connectivity • Problem Formulation • Maximize number of sleeping nodes so that convex region A is Ks covered and all active nodes are connected • Assumptions • Sensing and communication regions are circular • All nodes have same ranges, Rs and Rc • Nodes can communicate if they are within Rc • Nodes ensure sensing within Rs • Each node knows its location • Intuition: Ratio Rc/Rs is critical to coverage and connectivity

4. Problem continued • Connectivity requirement: any active node must be within Rc of at least one other active node • Coverage requirement: all locations in the coverage region must be within Rs of at least one active node

5. Problem continued • Coverage degree Ks and connectivity degree Kc • Can connectivity --> coverage? • No, connectivity only cares about active nodes • Can coverage --> connectivity? • Yes only if Rc >= 2Rs • Theorem: a set of nodes that Ks-cover a convex region forms a Ks-connected communication graph if Rc >= 2Rs

6. Coverage configuration protocol (CCP) • Configuration protocol based on Theorem above • Coverage --> connectivity when Rc >=2Rs • Can configure a network to a requested coverage degree • Decentralized protocol only requires local neighbor states to determine specific node states (eligibility rules) • All we need to do is to activate nodes to cover a region with a specified degree and we are connected! • Issues: how does a node know its sensing region is already covered?

7. Sufficient condition for coverage • Theorem: a convex region A is Ks-covered if all intersection points between nodes, and between nodes and the region boundary are Ks-covered • Intuitively all points in same patch have same coverage degree • Only checking the intersection points

8. Ks-coverage eligibility algorithm in CCP • A node is eligible to turn active if any intersection point inside its sensing circle is not already Ks-covered • steps • All nodes are active when deployed • Active nodes periodically broadcast/receive beacon messages • Sleeping nodes periodically wake up to receive these beacons and decide whether to stay waked up or not

9. What about the case when Rc < 2Rs? • CCP is not sufficient as coverage will not lead to connectivity • Solution: combine a coverage protocol CCP with a connectivity protocol SPAN • What is SPAN? • Connectivity-only protocol • Eligibility rule: a sleeping node will be eligible to be active if a pair of neighbors are unable to reach each other directly or through one or two active nodes • A sleeping node is within the communication range of an active node • Use both eligibility rules • Activate if either CCP or SPAN’s eligibility rule is satisfied • Sleep if neither is satisfied. • Hello beacons contain info about other active nodes