The project consist of a
theoretical and a system research thrusts in energy-harvesting wireless sensor
networks. Looking at the energy harvesting module in sensor node from the
system’s prospective, we observe that adopting an interruption/harvesting
policy enhances the energy consumption, but it also increases average packet
end-to-end delay and packet dropping! There are various tradeoffs exist in
wireless sensor network (WSN) design. Of particular interest to the project are
1) End-to-end delay vs. energy harvesting & network size, and 2) Increasing
network size allows for a smaller number of data sink nodes and reduces
dropping but it also increases average end-to-end delay. In order to quantify
the tradeoffs, we raise a question. How delay, network size, and harvesting
policy (service vacation) interact with each other? In other words, how large
the network dimensions can go to considering certain packet latency threshold
and dropping? Energy harvesting module can arbitrary be triggered, upon empty
buffers, and thus, imposes random interruption periods on the sensor node,
these random cycles (vacations) increases latency due to residual vacation time
that is consumed for harvesting. In order to solve the above challenge, the
student will study this system and think of a theoretical and a system
approaches (with the assistance of the instructor) to improve the end-to-end
delay. For instance, a possible solution would be, instead of arbitrary
harvesting time, we aim at optimizing this value to minimize the delay and
dropping. We also aim at adding a constraint that forces the sensor node to
trigger the harvesting phase once its battery is low! Currently, the networking
lab has a set of 20 sensor motes that are programmable using TinyOS and also
has all the necessary mathematical packages to evaluate the student proposed
models.

The
student is expected to achieve:

1. (theatrical component): An optimization of the energy harvesting value that minimizes the system end-to-end delay validated using matlab.

2. (systems component): Translate the results from (1) into a working code, which can be tested over our TelosB sensor motes using TinyOS (C/C++).

The above results, if completed, are considered novel and can result into a publication with the agreement with the instructor.

Computer science and electrical engineering