Aside from their influence on the maximum trackable speed, the leader heartbeat period and the awareness horizon also significantly influence message related energy consumption. For different hardware this effect will vary. However, the fundamental trends remain consistent.
Figure 8 shows energy consumption for varying the leader heartbeat period and awareness horizons explored in Figure 7. The effect of the heartbeat period on energy consumption is significant. An interesting point to notice, however, is that the alternative (horizon, period) tuples that track a particular speed consume almost the same amount of energy. Following up on the example from the previous section, the three alternative parameter tuples that track a speed of 50 m/s, namely, (1 hop, 1600 ms), (2 hops, 2900 ms), and (3 hops, 4200 ms), consume roughly the same energy of 4 units. This is seen by finding the intersections of the vertical lines at 1600 ms, 2900 ms and 4200 ms with the energy curves for the corresponding horizon. It therefore appears that choosing a larger horizon does not have an advantage as long as the heartbeat period is appropriately chosen. This is intuitive. Increasing the awareness horizon allows using a larger heartbeat period. However, it also increases the number of messages sent, since each heartbeat is flooded in a larger radius. The net sum of messages exchanged, therefore, remains largely unaffected.
In view of the above, an argument can be made for smaller horizons, since they reduce the number of nodes inhibited from creating new entities, thus making the sensor network more responsive to the advent of new events into the environment. Nodes in a network with a large awareness horizon will attribute measurements of such new events to the entities they are already aware of; an effect that should be avoided.
The set of experiments presented above illustrate the ability of our architecture to track events at varied speeds in a sufficiently dense sensor network. A system designer can employ this architecture, choosing the leader heartbeat period and the awareness horizon in accordance with the expected speed of migrating events, available bandwidth of the wireless medium, and energy restrictions (required system lifetime) of the hardware being deployed. An implementation of our architecture on the MICA motes which is discussed in the following section.
