research
My research interests focus on the fields of Communications and
Signal Processing with more emfasis in the following areas:
- Sensor Networks
- Dynamic Resource Allocation
- Adaptive Algorithms
- Coding Theory & Techniques
- Channel Estimation / Equalization
- Embedded Systems
PhD Research
Advisor: Prof. Urbashi Mitra
I am currently working on the KNOWME Networks project under the guidance of my advisor, Dr. Urbashi Mitra. The focus of this project is efficient physical activity recognition by using a network of heterogeneous sensor nodes for health– monitoring applications. Specifically, the KNOWME project team has designed a system consisting of a set of heterogeneous sensors such as accelerometers and heart-rate monitors and a cellphone fusion center. This Wireless Body Area Network (WBAN) is on-body and the sensors measure vital signs such as heart-rate, oxygen levels, etc which in turns communicate to the fusion center via Bluetooth. The generated measurements are then used to estimate the current activity performed by an individual from a set of predefined activities such as standing, walking, sitting, runnning etc. However, it has been observed through experimentation that continuously collecting samples from the sensor nodes using the Bluetooth protocol consumes significant amount of energy which leads to limited network lifetime since it drains quickly the battery life of the fusion center. In other words, the fusion center is the energy bottleneck in this case in contrast to traditional sensor networks, in which the sensor nodes themselves are energy constrained.
As far as my contribution to this project is concerned, I work on the design and implementation of sleep scheduling techniques (aka optimal sensor selection policies) using the theory and tools of stochastic control and optimization. My research efforts focus on modeling the system of interest as a Partially Observable Markov Decision Process (POMDP) and deriving optimal control laws that will prolong the system's lifetime. In addition, I am actively investigating low-cost approximation schemes that can be implemented and tested in the KNOWME project. The above problems are very challenging and interesting not only due to the heterogeneous nature of the sensors, a fact that implies different discriminating capabilities and power consumption, but also due to the limited energy budget of the fusion center.
Diploma Thesis Research
Advisor: Prof. Kostas Berberidis
One of the most important problems, especially in wireless communications, is the estimation of the channel between the transmitter and receiver. Several approaches have been proposed in the literature with the most common to use a training sequence, which is transmitted between the information symbols. However, this leads to bandwidth waste. Therefore, novel techniques such as blind and semi – blind channel estimation algorithms have been proposed so as to improve performance.
In my diploma thesis, I studied a more recent approach according to which the training sequence is a periodic sequence that is superimposed on the information symbols. I implemented several channel estimation algorithms of this kind and I compared their performance under different scenarios. Finally, I identified future directions and open problems regarding this promising family of channel estimation algorithms.
