B. Software Modules
1) Classification algorithm running on the STP: A low
complexity classification algorithm based on energy and correlation
features is implemented on a Wismote running Contiki
Operating System (OS) as described in [3]. A two-step classification
process is adopted. In the first step, the algorithm
classifies between intruder and clutter. The detected intrusions
are further classified as being either a human or an animal.
2) Wireless Communication: The STP, relay nodes and
border router communicate using IPv6 over a Low-power
Personal Area Network (6LoWPAN) that runs on top of IEEE
802:15:4. There may be packet drops on account of the nature
of the wireless medium. Multiple copies of the same decision
are transmitted to the border router to combat packet loss.
For each classification decision, three packets are sent from
the STP to the border router. To ensure that packet drops do
not pass undetected, each packet carries as part of its data, a
sequence number for the decision. Missing sequence number
for a decision on the GUI represents that all three packets
corresponding to that decision got dropped.
3) Time Synchronization for continuous video recording:
The webcam should be time synchronized with the server to
detect if any intrusions are missed by the STP. To determine the
number of missed detections, the continuous video recorded