3.5.1 The Logic of Aggregation: Communication vs. Computation

In a WSN, the energy required to transmit 1 bit is equivalent to executing 1000 to 2000 CPU instructions. Therefore, the engineering goal is to process data locally and send only meaningful summaries.

3.5.2 In-Network Aggregation Techniques

1. Tree-based Aggregation: Intermediate nodes compute summaries (SUM, AVG, MIN, MAX) as data travels up the RPL tree. Reduces traffic near the root.
2. Cluster-based Aggregation: Data is normalized, compressed, and encrypted at the Cluster Head before backhaul.
3. Temporal Aggregation: A node averaging 10 readings over time and sending only 1 packet.
4. Spatial Aggregation: Neighbors compare readings; only the highest value node sends the data (used in fire detection).

3.5.3 Data Dissemination (The Information Push)

Spreading configuration data or firmware from the gateway to all nodes.

• SPIN (Sensor Protocols for Information via Negotiation): Uses Metadata (ADV) to query if a node already has the data before sending it, avoiding "Broadcast Storms".
• Trickle Algorithm: Ensures data consistency across the network by broadcasting only when a discrepancy is detected (used in RPL). It uses exponential timers to keep the network silent when stable.

3.5.4 Information Fusion Theory

Combining heterogeneous data (e.g., Sound + Vibration + CMOS Image) to increase decision confidence.

• Case Study: A vibration sensor triggers an alert. A camera node wakes up to verify. The fusion reduces false alarms by 90% and saves cloud bandwidth.