Foto del docente

Flavio Zabini

Senior assistant professor (fixed-term)

Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi"

Academic discipline: ING-INF/03 Telecommunications


Keywords: Echo Canceller Partial Equalization Random Sampling Point Processes Performance-Fairness trade-off

1. Multidimensional random sampling for crowd-sourcing

2. Partial equalization in Multi-carrier CDMA.

3. Echo canceller for DVB-T repeaters.

4. Performance-fairness tradeoff in communication systems


1. Multidimensional Random sampling for crowd-sensing

Multidimensional reconstruction of signals is a key enabler for emerging applications including array signal processing, magnetic resonance imaging, seismology, digital communication and control, software defined radio and networks, vehicular networks, and environmental monitoring. Big data and crowdsensing applications can be associated with multidimensional random sampling (e.g., to reconstruct spatial distribution of data).
In particular, various applications in sensor networks imply a signal reconstruction entity collecting samples from sensors randomly scattered in an environment. The presence of signal sources scattered according to a homogeneous Poisson Point Process is also common in recent works on wireless communication and localization networks. However, homogeneous Poisson point processes do not always accurately describe the sample spatial distribution in many cases of interest (e.g., sensors can be scattered according to different densities in regions of a monitored area, or according to a point process model with repulsion/attraction between points). Moreover, in real scenarios there might be uncertainties due to sample losses, imperfect measurements, and imperfect knowledge of sample location information. Such uncertainties can be detrimental for signal reconstruction and call, together with a sample spatial distributionw with inhomogeneity and stochastic interaction between points, for a new methodology to analyze multidimensional random sampling. My current research activity is thus aiming to merge classic sampling theory with modern spatial point processes theory.


2. Echo canceller for DVB-T repeaters.

One of the advantage in using OFDM signals in broadcasting is the possibility to employ single frequency networks. Unfortunately, in the on-channel repeaters, the presence of a coupling channel between the transmitting and the receiving antennas gives origin to one or more echoes, having detrimental effects on the quality of the repeated signal and critically affecting the overall system stability. My research focused on a low-complexity echo canceller unit able to counteract the positive feedback of the coupling channel through an adaptive finite impulse response filter in negative feedback which reproduces the coupling channel itself (after having estimated it through local low-power training signals). My contribution include an analytical study about the stability and the Mean Rejection Ratio.


3. Partial equalization in Multi-carrier CDMA.

Multi-carrier code division multiple access can support high data rates in next-generation multi-user wireless communication systems. Partial equalization (PE) is a low-complexity technique for combining the signals of sub-carriers to improve the achievable performance of MC-CDMA systems in terms of their bit error probability and bit error outage with respect to well-known maximal ratio combining (MRC), orthogonality restoring combining (ORC), and equal-gain combining (EGC) techniques. In the presence of fading channels, the orthogonality between sequences of different users is destroyed even in the downlink. Therefore, multiple access interference has to be taken into account. As known, in a noise limited situation, MRC is the best choice, while ORC results optimal in an interference-limited situation. EGC represents a middle way. Partial Equalization represents the best choice in all the intermediate situation, keeping the receiver complexity low as MRC, ORC and EGC. My contribution consisted in having found the analytical expression of the optimum PE parameter as a function of the number of subcarriers, number of active users (i.e., the system load), mean signal-to-noise ratio (SNR), and variance of the channel-estimation errors for the aforementioned block-fading channel. Previously, it was known only by simulation and only in not realistic fading channel situation. Realistic imperfect channel estimation and frequency-domain (FD) block-fading channels are considered. More explicitly,  I showed that the choice of the optimal PE technique significantly increases the achievable system load for the given target BEP and BEO.

4. Performance-fairness tradeoff in communication systems

Optimal resource allocation is an outstanding issue in wireless communication systems. My research has focused on the ever challenging optimization of the tradeoff between throughput and fairness. Specifically, a novel framework has been proposed to analytically derive the maximum average throughput versus fairness under the assumptions that the throughput of each user i) increases if more resources are allocated to him and ii) depends on how many resources and not which resources are allocated. A general formulation of the tradeoff optimization problem is achieved and a closed form solution is derived and validated in those scenarios where throughput linearly depends on resources, which cover several realistic cases. The case of nonlinear relationship is under investigation.