ANALYSIS OF WILDCENSE FOR HABITAT MONITORING
daiict · 2012
Pratik Sajnani (ID: 200801059) — Mentor: Prof. Sanjay Srivastava
src_origin: Dhirubhai Ambani Institute of Information Technology (DA-IICT) // year: 2012
This thesis analyzes the wildCENSE protocol for habitat monitoring applications in mobile wireless sensor networks. Sensors placed on animals form an ad-hoc mobile network, relaying tracking data to a centralized base station. We design and evaluate a 5-phase communication protocol covering GPS-based synchronization, peer discovery, pre-processing, data exchange, and post-processing. The protocol is simulated using NS2/OTcl and analyzed for efficiency, reliability, and energy consumption. Edge cases including GPS failure recovery and isolated node handling are addressed.
Wireless sensor networks (WSNs) are purpose-built networks for monitoring and tracking physical phenomena. Nodes are sensors placed on or around objects of interest. In the case of mobile objects — animals in their natural habitat — the sensors are also mobile, placing strict constraints on energy utilization and memory capacity.
1.1 Sensor Architecture
Each sensor node in the wildCENSE network is composed of five subsystems:
- 1.Sensing device — tracks the host animal via accelerometers, temperature sensors, or GPS receivers
- 2.Micro-controller — overall integration, protocol execution, and power management
- 3.Memory (non-volatile) — temporary local storage for buffered readings
- 4.Transceivers — transmit and relay data between nodes over short-range radio
- 5.Interfaces (GPS, serial ports) — positional data and external communication
[Figure] Sensor node block diagram. Five subsystems (sensing, micro-controller, memory, transceiver, interfaces) connected via internal bus. Power management unit regulates battery drain during active and sleep duty cycles.
1.2 Sensing vs. Sending
Each node performs two primary functions: collecting data (sensing) and transmitting it (sending). Sensing is usually hardware-dependent, operates within a limited signal range, and probes at regular configurable intervals.
When sensors are mobile and the destination is a stationary base station, data must be relayed through a multi-hop network. The network itself becomes a critical resource. Key concerns: efficiency (probability of successful delivery), reliability (probability data reaches the base station), energy, bandwidth, and computing power.
The habitat monitoring problem: sensors are placed on animals and must function autonomously for a year or more, periodically sending accumulated tracking data to a centralized base station. Animals move freely; encounters between sensor-carrying animals are opportunistic. The sensor network must exploit these encounters to relay data efficiently.
Sensors form a cooperative mobile ad-hoc network. When two animal-carried sensors come within radio range of each other, they exchange data that the other node lacks — propagating information toward the base station in a store-and-forward fashion. We require a protocol that governs this communication.
[Figure] Habitat monitoring topology. Animals (A1\u2013A6) carry sensor nodes. Arrows show opportunistic data relay paths. Base station (BS) receives aggregated data whenever any node comes within direct range or through multi-hop relay.
The wildCENSE protocol consists of five phases executed whenever two nodes enter radio range of each other. Each phase is designed to minimize redundant transmissions and energy consumption while maximizing the probability that data reaches the base station.
3.1 Phase 1 — Synchronization
Nodes operate on internal clocks with configurable duty cycles (active/sleep periods). Before any data exchange, clocks must be aligned so active windows overlap. GPS receivers provide timestamps used as a universal time reference.
- ·Primary sync: GPS timestamp comparison between two nodes on first contact
- ·Drift correction: periodic re-synchronization to prevent accumulated clock drift
- ·GPS failure fallback: if GPS is unavailable, the node with the more recent known-good timestamp is treated as the reference
- ·Duty cycle negotiation: nodes agree on a shared active window for the duration of the encounter
3.2 Phase 2 — Peer Discovery
Before data exchange, nodes must identify their neighbors and determine what data each node holds. This is accomplished via probing messages. Two discovery schemes are defined:
- 1.Contention-free: each node broadcasts probe messages at predetermined fixed or random time slots. No collision handling required; slots are allocated to avoid overlap.
- 2.Contention-based: nodes compete for the channel using CSMA-style back-off. The first node to acquire the channel broadcasts; others wait. Suitable when time-slot coordination is infeasible.
The probe message contains the node ID, current timestamp, and a compact summary of the data the node holds. This summary is used in the pre-processing phase to determine what is actually worth transmitting.
3.3 Phase 3 — Pre-Processing
Naive data exchange would transmit all buffered readings — most of which the other node likely already has. Pre-processing eliminates redundant transmissions by comparing the data inventories of both nodes before any bulk data transfer begins.
Each node maintains three data structures per peer:
- ·Dump bits — a bit-vector indicating which time-slotted data records the node holds
- ·Popularity numbers — a count of how many nodes in the network are known to hold each record (estimated from probe messages)
- ·Send bits — derived from dump bits and popularity numbers; flags records the other node does not have and that are rare (low popularity)
Control packets exchange dump bits between both nodes. Each node computes the send bits for its peer: records it holds that the peer lacks. Only send-bit-flagged records are transmitted in the data exchange phase.
[Figure] Pre-processing data flow. Node A sends its dump-bit vector to Node B. Node B computes send-bits = (A_dump XOR B_dump) AND A_dump, identifying records A has that B lacks. The same computation runs symmetrically.
3.4 Phase 4 — Data Exchange
Nodes exchange records identified by the send-bit computation in Phase 3. Two sub-schemes govern the exchange:
- 1.Round-robin: nodes alternate transmitting records in time-slotted fashion, ensuring neither node monopolizes the channel during a short encounter window
- 2.Acknowledgement-based: each transmitted record requires an explicit ACK from the receiver before the next record is sent; retransmission on timeout
When a node's data reaches the base station — either directly or via relay — that data is freed from local memory. The base station broadcasts periodic "flush acknowledgements" listing record IDs it has received, allowing all nodes that relayed those records to reclaim memory.
3.5 Phase 5 — Post-Processing
After data exchange, nodes perform cleanup to prevent stale or duplicate records from persisting in the network. Post-processing uses timestamp-based redundancy checks: records older than a configurable threshold that have already been widely propagated (high popularity number) are candidates for eviction from local memory.
4.1 GPS Failure Recovery
If a node's GPS receiver fails, it cannot provide a reliable timestamp for synchronization. The recovery procedure:
- 1.Node broadcasts GPS-failure flag in probe messages
- 2.Peer nodes share their current GPS timestamp with the failed node
- 3.Failed node uses the most recent received timestamp as its reference clock
- 4.If multiple peers are available, the median timestamp is used to resist outliers
- 5.Recovery is re-attempted on the next GPS acquisition cycle
4.2 Isolated Node Handling
A node that has not encountered any peer within 5 duty cycles is considered isolated. Standard send-bit-filtered exchange would suppress most of its data (since it has no knowledge of what the network holds). An isolated node instead performs a full data dump to the first peer it encounters, regardless of popularity numbers. This ensures data is not lost due to prolonged isolation.
The wildCENSE protocol was implemented and simulated using NS2 (Network Simulator 2) with OTcl scripting. Simulations modeled varying network densities, encounter rates, and data sizes to evaluate protocol performance under realistic habitat conditions.
5.1 Simulation Parameters
- ·Network size: 10–50 nodes
- ·Mobility model: random waypoint (approximating animal movement patterns)
- ·Radio range: 30–100 meters (depending on terrain assumptions)
- ·Data generation rate: 1 record per duty cycle per node
- ·Duty cycle: 60 seconds active, 300 seconds sleep
- ·Simulation duration: 24 simulated hours per run
5.2 Metrics
- ·Delivery ratio: fraction of generated records that reach the base station
- ·Redundant transmissions: transmissions of records the receiver already holds
- ·Energy consumption: estimated from active radio time and processing cycles
- ·Memory utilization: average buffer fill level across all nodes
[Figure] NS2 simulation output. Top graph: delivery ratio vs. node density for contention-free and contention-based discovery schemes. Bottom graph: redundant transmission rate with and without pre-processing. Pre-processing reduces redundancy by ~60% at medium densities.
The second component of this thesis investigates topology formation in mobile ad-hoc networks (MANETs) with application to video conferencing quality of service (QoS). In both sparse and dense topologies, uncontrolled flooding creates excessive traffic that degrades call quality.
The proposed protocol improves QoS by organizing nodes into groups. A group leader aggregates traffic from members and relays it via a designated relay sub-group. This hierarchical structure reduces the number of active transmitters at any given time, lowering collision rates and improving end-to-end latency for video streams.
- ·Group formation: nodes cluster based on signal strength and mobility prediction
- ·Leader election: node with highest stability score (fewest recent handoffs) becomes leader
- ·Relay sub-group: two or three high-connectivity nodes bridge between groups
- ·Dynamic re-clustering: triggered when group membership changes by more than 20%
The wildCENSE protocol provides a structured, energy-efficient approach to data collection in mobile wireless sensor networks for habitat monitoring. The five-phase design — synchronization, peer discovery, pre-processing, data exchange, and post-processing — minimizes redundant transmissions while maintaining high delivery ratios even with sparse node encounters.
Pre-processing with dump bits, popularity numbers, and send bits is the key innovation: it compresses the coordination overhead into compact control packets and eliminates the bulk of redundant data transfers. GPS-based synchronization and isolated node fallback procedures make the protocol robust to hardware failures and long separation periods.
The MANET topology work demonstrates that group formation with relay sub-groups can substantially reduce collision rates in both sparse and dense deployments, with direct application to video conferencing QoS in ad-hoc environments.