The Entity Communication Problem

The problem addressed by our architecture is more formally described as follows. We consider a dynamically changing set $S$ of events in the physical environment of the sensor network. Let the physical location of each event $E_i \in S$ at time $t$ be denoted $L_i(t)$. A node is said to be in the vicinity of event $E_i$ at time $t$ if it is within sensor range of the event's location, $L_i(t)$. In this paper, we assume that environmental events are localized. In other words, their location is described by a single point in space, as opposed to an area. This definition applies to tracking vehicles, finding survivors, monitoring wild animals, or detecting localized fires. It does not apply to applications involving distributed phenomena such as detection of large chemical spills. We assume that events can be detected independently by individual nodes in the sensor network based on their local measurements. For example, detecting a magnetic signature in a desert battle area would usually be indicative of a passing armored vehicle. Finally, we assume that events are sparse. In other words, the signatures of different targets are generally not overlapping.

Let $T_i$ denote the set of nodes in the vicinity of event $E_i$. The objective of our architecture is to maintain a unique addressable destination associated with each event $E_i$, such that sending data to this logical event address causes delivery of this data to $T_i$ regardless of the location $L_i(t)$. In the current implementation, we elect a leader out of set $T_i$. The leader, among other things, is responsible for communication with remote destinations. Hence, in the above problem statement, we define delivery of a message to $T_i$ as delivery of the message to the current entity leader who by definition belongs to the set $T_i$. What the leader does with the message is an orthogonal issue in our architecture.

Note that once the aforementioned addressing and communication problem is solved, it becomes trivial to associate multiple communication end-points with each entity simply by demultiplexing the received message based on a port number in the message header, in the same sense that UDP creates multiple ports over IP.