Scalable and Adaptive Polling Protocol for ConcurrentWireless Sensor Data Flows

The uncoordinated transmission of sensor measure- ments to a collector over simple wireless technologies can lead to severely degraded performance when operating close to channel capacity. In this case, it can be far more effective to let the collector poll the sensor nodes to retrieve their data. We propose a coordinated protocol that exploits the broadcast ca- pability of the wireless channel to eliminate contention between nodes and to minimize the number of packet transmissions performed at each poll. Furthermore, to let the protocol follow the dynamic changes in the sensor data generation rate, we propose an application-agnostic method for estimating the data generation rate of each node locally, and extend the protocol so that the collector dynamically adjusts the polling rate accordingly. Our method is based on a Kalman filter tuned for stable data generation rates, which is controlled via simple signals generated at runtime by the underlying polling protocol. We experimentally evaluate an implementation of our protocol for different data traffic scenarios using real nodes that communicate with the collector over IEEE 802.15.4 radio, showing that the collector can successfully track the actual data rate for both periodic and stochastic data generation.