In today’s world, we are surrounded by portable wireless devices, each undertaking a single or multiple signal processing tasks.
HANDiCAMS aims at developing a new paradigm in which heterogeneous devices with different device-specific tasks cooperate to achieve superior performance.
The project is aimed at developing a new ICT paradigm where multiple heterogeneous devices that cooperate in multiple signal processing tasks, which is referred to as Multiple Devices for Multiple Tasks (MDMT). This is radically different from current ICT paradigms, in which stand-alone devices merely focus on individual tasks or multiple devices perform one joint task, e.g., in a Wireless Sensor Network (WSN). Examples of the heterogeneous devices considered are tablets, smartphones, handheld cameras, active headsets and hearing aids. Each device is equipped with one or several sensors, e.g., microphones and cameras, as well as with computing and wireless communication facilities, and has its own signal processing task, e.g., a local signal enhancement task. The aim is to achieve superior performance in these tasks through cooperation amongst the devices, which then effectively act as nodes in a WSN type set-up, where each node contributes to the other nodes’ tasks.
The main objective is to develop distributed, cooperative and adaptive signal processing algorithms for the acquisition, coding, processing, and in-network fusion of multimedia signals, in particular for the enhancement of audio and video signals. The algorithms are operated in a heterogeneous, ad-hoc and dynamic network, where each node has its own signal processing task as well as its own specific mode of operation. Furthermore, the algorithms should be scalable and require minimal communication bandwidth and power. As the network nodes may be selfish or opportunistic, general operating principles will be designed that provide incentives for cooperation. A general bottom-up design strategy will be adopted, rather than the usual top-down approach used in WSNs.
The project will yield new theoretical frameworks for distributed detection, classification, estimation, coding, topology inference and cooperation strategies. In addition, two use cases are proposed in the context of audio and video enhancement, which will eventually serve as a proof of concept.
HANDiCAMS starting date: October 1, 2013