Conclusions

The purpose of this thesis was to implement the FM-WiFIX proposal on a real testbed, in order to perform a proof-of-concept of this solution. With this implementation, it was expected to address the three main problems of the Wireless Video Sensor Networks (WVSNs): throughput unfairness, bad performance and energy inefficiency.

The key concept on which this solution relies is the low power FM-RDS based out-of-band control channel. However, this solution is a rather ambitious one, that proposes the utilization of this out-of-band control channel in a way that, to the best of our knowledge, was never used before: signaling the transitions between on and off states of the Wi-Fi card in a matter of milliseconds.

Throughout our work, we have faced several challenges: the original FM-WiFIX had some problems that had to be solved to enable a successful implementation; the RDS technology was not the best choice to implement the control channel of the FM-WiFIX solution, which must be reliable, fast and provide a great coverage; the absence of support to effectively turn on and off the Wi-Fi radios; and the lack of processing power of the Raspberry Pi platform, that limited our network’s performance.

Nevertheless, recurring to some turn around solutions and tweaks, we were able to successfully deploy a proof of concept prototype of a WVSN, implementing the FM-WiFIX proposal.

 Contributions

With this thesis we have achieved many important contributions. We suggested and implemented multiple modifications to improve the FM-WiFIX concept, regardless of the technology used for the control channel (the control protocol, the register phase, the polling order and several others); we have characterized and adapt the Radio Data System to serve the purposes of the proposed control channel, although we have also noticed that the FM-RDS is not the best solution for the FM-WiFIX paradigm; we have designed a new scheduling mechanism for FM-WiFIX (the transmission burst scheme) that enables its operation with the limitations imposed by the RDS channel; we have characterized the power consumptions of the Raspberry Pi and the associated hardware; conducted a study about a multitude of paths to quickly turn on and off the Wi-Fi NIC; and finally, we concluded this work with an evaluation of the FM-WiFIX solution in several contexts, that effectively was this system’s proof-of-concept as this evaluation has validated the simulation results obtained in [1].

Future Work

The main improvements that should be performed, in order to unlock the full potential of the FM-WiFIX solution and to achieve an optimized WVSN, for real time video streaming, are the following:

• It is mandatory to choose a new technology for the control channel. RDS is too slow, too unreliable and only achieves acceptable polling rates if we compromise its range. The Nordic’s nRF24L01 may be a good solution.
• From vital importance is also the identification and assembly of a new testbed, without using the Raspberry Pi model B. As previously described, this platform had become the bottleneck of our solution.
• Find a solution that allows the effective transition between on and off states of the Wireless NIC, without compromising the association to the network.
• As optimizations, we considered some options as frame aggregation throughout each branch, using jumbo frames and spatial reuse to improve the system’s throughput. Although we choose to spend our time studying other aspects of the solution, these are indeed very interesting paths of research to follow, in order to improve even more the FM-WiFIX solution.