Advanced, hardware supported in-network processing for the internet of things

We present a novel wireless sensor network (WSN) architecture that relies heavily on the in-network processing of information. The information captured by the WSN nodes is being filtered and transformed by other nodes so that fewer pieces of information of higher semantic value reach the network sink and are fed to the applications. This type of IoT infrastructure assists in the efficient energy management within the network (and the elongation of network lifetime) and the reduction of the processing overhead incurred by the application. The application has to "confront" a small volume of data with the very specific focus that it requires. This in-network processing architecture usually involves information filtering/transformation nodes that rely on computationally demanding algorithms. Such nodes are based on accelerated, not conventional hardware that expedites all the advanced processing with a minimum energy overhead. This is imperative for the efficient operation of the WSN as it addresses two important needs: performing demanding tasks at low energy cost, filtering the information to support energy efficiency and render the network sustainable over time.