Prof. Leshem's Lab
Head - Signal Processing, Imaging, Radio Communications & Astronomy Lab
Personal website: http://www.eng.biu.ac.il/leshema/
Group website: https://www.eng.biu.ac.il/spiral/
Prof. Amir Leshem is a full professor in the Faculty of Engineering, where he heads SPIRAL (Signal Processing, Information networks, Radio communications and Astronomy Lab). The lab’s research falls into several main research areas, signal processing, data analysis, communications and game theory, and their applications to sensor networks, social networks, radio astronomy and future cellular (5G) and cognitive ad-hoc radio networks.
Leshem and his team work to reconstruct signals from noisy measurements using a variety of statistical techniques. They conduct research on both sensor networks and arrays of sensors (groups of connected sensors that do not have limitations on bandwidth, power, connectivity or energy). Recently the group is also developing a social radar capable of actively mapping the influence structure of social networks.
In sensor networks, they work on distributed processing and attempt to perform optimal signal processing within the given limitations of communication, energy, and bandwidth. Applications of their work on sensor arrays include radio astronomy imaging, wherein images of the sky can be obtained using measurements from radio telescopes. Leshem's techniques served as the basis for the calibration of the recently built Low Frequency Array and are likely to influence the Square Kilometer Array.
Communications and networking
Leshem and his group focus on cross-layer design for communication and data processing networks. One of the most interesting directions developed by the group is the exploitation of standard medium access techniques to solve, distributedly, hard combinatorial resource allocation as well as energy minimization problems in networks. This line of research combines deep game theoretic techniques with system level implementations of the algorithms.
The group also studies the related problem of improving data rates for next-generation DSL networks. Since each DSL is a separate line, it is susceptible to interference from other lines and therefore transmits data at a lower data rate. In 2003 Leshem and his collaborators were the first to demonstrate symmetric transmission rates of 100 Mbps over telephone lines to distances up to 500m. At that time they already recognized that it would be possible to push DSL to rates of 1 Gigabit per second. Indeed, the recent g.fast standard realizes these insights. The group is still involved in developing efficient communication techniques for wireline communications.
Spectrum allocation is an important problem for the efficient use of the electromagnetic spectrum. With the booming increase in mobile internet as well as the forthcoming Internet of Things this problem will increase. The group is developing novel optimal spectrum allocations techniques that are computationally efficient and distributed. This reduces the communication overhead for network management.