Auto-configuration In Wireless Sensor Networks Nakjung Choi, Haeyong Kim, Minji Nam April Fools’ Day

Our team Members – Nakjung Choi, Minji Nam, Haeyong Kim Rough Ideas – Node identification? Use information of node position? (low mobility) Routing protocol uses the information? Ex) auto-assign prefix – Auto-config + topology management? Coverage area in wireless sensor networks Node’s sensing coverage, transmission coverage Other factors Calculate appropriate node density? –If possible, construct topology → grid size, address field size)

A Self-Organizing Approach to Data Forwarding in Large-Scale Sensor Networks

Introduction Frequent sensor failures, stringent energy constraints – Unique design challenges for data forwarding Sensor Networks – Stimuli and sinks – Very susceptible to sensor failures Energy depletion or destruction In this paper – Self-optimizing multicast tree-based forwarding infra. – Node failure recovery algorithm

Design Issues Design issues for data forwarding protocol – Scalability No global knowledge, no global synchronization – Reliable delivery In spite of unreliable individual sensors – Limited resources Limited energy supply, mem. Size, processing pw. – Error-prone wireless medium Lower bandwidth – No globally unique ID Cannot rely on the existence of a unique receiver

The Multicast Tree Protocol The minimum spanning tree – Infeasible, so through local mechanism Basic operation – Through sinks’ advertisement process, construct multicast tree – In the absence of sink interests, sensors are inactive – In the absence of stimuli, low-frequency hello message

The Multicast Tree Protocol First phase – Broadcasting sink advertisements – SinkAdvertisement message SinkID and nofhops By setting a TTL value Second phase – Election of candidate merge points SinkID and nofhops in Sink Table – If nofhops is less than TTL value, forward Third phase – Election of one merge point – MergeAdvertisement The sum of distances from itself to all sinks A set of their SinkID

The Multicast Tree Protocol Fourth phase – Branch formation towards the sinks – MergeAcknowledgment towards all sinks in its sink table Builds branches of the tree – All neighbors of a sensor have different IDs Maintaining the tree – KeepAlive message at very low frequency Forming a new branch – Joining of the new node – Elect the sensor on the tree that is closest to new node – MergeAcknowledgment towards the new node

The Multicast Tree Protocol Report forwarding – Receive from its upstream neighbors and forwards it to its neighbors on the tree – Only upstream and downstream IDs in routing tables Tree damage and reconnection – Detection of sudden node failure Using timer and passive acknowledgment – For smooth handoff from a low-energy sensor Informs its upstream and downstream neighbors of its failure

The Multicast Tree Protocol

Performance Evaluation Using Parsec RPF (Reverse Path Forwarding) TRPF (Truncated RPF) MPIM (Merge Point initiated Multicast) SIM (Sink Initiated Multicast) RPF – large overhead TRPF – comparable to the multicast-based approach, but large storage requirement MPIM is slightly better than SIM

References [1] Mirkovic, J.; Venkataramani, G.P.; Lu, S., Zhang, L., “A self- organizing approach to data forwarding in large-scale sensor networks,“ ICC IEEE International Conference, June 2001 [2] Alberto Cerpa and Deborah Estrin, “ASCENT: Adaptive Self- Configuring Sensor Networks Topologies,” in Proceedings of the Twenty First International Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM 2002), June [3] “A Two-Tier Data Dissemination Model for Large-scale Wireless Sensor Networks”