SEA Lab has a strong commitment to developing low-cost machinery that can support emerging intelligent methods and, at the same time, limit costs to ensure maximal diffusion of the related technologies.
SEA Lab's research on embedded system design strictly integrates with the other research areas, as embedded electronics is always seen as the target environment for practical and industrial applications of the various methods and technologies.
The main research areas in this context comprehend:
Cryptographic algorithms and protocols that are
believed secure in theory are often liable to breaks when implemented in
practice. This is mainly due to coding flaws or to subtleties in computational
aspects. The research at SEA Lab aims at developing secure implementations of
existing algorithms and protocols on embedded electronics architectures.
Target platforms typically envision both FPGA-based and DSP-based systems. This made it possible also to perform some comparative approach between the two platform alternatives. Some significant results and lines of research in this area include:
Today's challenges in Logistics mainly involve the continuous monitoring of transportation systems, which is progressively moving from a centralized approach (one center where all information flows to) to a distributed architecture, in which all agents carry some information-generation ability and possibly limited computing power. This model is consistent with the "Ambient Intelligence" paradigm that represents one of the most promising directives in the next decades' research.
Embedded distributed systems envisions hardware machinery that is hosted by individual subjects or agents (e.g., vehicles, trucks, train car, etc.) and carries some limited computing ability. This endows each agent to have and disseminate useful, local information such as current position, traffic status, routing abilities, etc. In addition, smarter versions of the overall set-up may include local-level intelligent algorithms for decision-making and optimization of some cost function (e.g., backlog, safety, etc.)
This set of activities strictly relates to the areas of Intelligent Algorithms for Video Surveillance and Visual Quality Evaluation.
SEA Lab's Research in this area pursues efficient embedded implementations of those intelligent algorithms for security and video surveillance. The basic electronic systems involve DSP boards with single/multiple color cameras.
The video-processing approaches have been tested under different DSP architectures, including components by Texas Instruments and Analog Devices.
In the video-surveillance application, the supporting platform currently adopted is the "Blackfin BF 533" by Analog Devices, which well suits the application requirements thanks to its hardware structure and multimedia-signal capabilities.
In the video quality-prediction research, the neural method has been tested on Texas Instruments TMSC320C6701 and TMSC320C6201 DSP devices.
In all cases, experimental results show that the hardware platforms can effectively support real-time processing for the related target applications.