DEEP LEARNING FOR MULTIPLE SENSORY DATA PROCESSING
The end-to-end experiment is given in a Matlab live script for the task of activity recognition using Deep Convolutional Neural Network in Multi-Sensor Systems.
PHYSICAL ACTIVITY RECOGNITION USING MACHINE LEARNING TECHNIQUES - [EXPERIMENT ON REALISTIC DATASET]
The end-to-end experiment is structured under a Matlab live script with detailed explanation and description for beginners.
Please access this site for download https://sites.google.com/site/thienhuynhthe/home
FIELD BASED TRAFFIC LOAD BALANCING FOR INDUSTRIAL WIRELESS SENSOR NETWORK
Traffic control and energy optimization is the goal of this paper. Sensors are usually distributed in random manner, which results in higher traffic at densely populated areas, and less traffic in areas of sparse density. Sink in the populated area spend more energy in gathering and processing data generated from sensors within its coverage area. This paper proposes a load balancing and routing algorithm suitable for industrial environment with a large number of sensor nodes and multi sinks. All the sink estimate amount of work each will do by measuring the volume of traffic and the area of each cluster. In a dynamic manner, the sinks will adjust their cluster sizes to ensure even distribution of work load. Sink in high traffic density areas will reduce their radius while the ones in less dense area will increase their radius. The nodes route data to sink by using Coulomb’s law. The forwarding of data packets from nodes to sink follow the direction of higher field gradient. Nodes at lower field transmit to the ones in higher field until the packet gets to sink. The simulation works shows how data traffic influences cluster size, and field intensity. Results show that proposed routing protocol performance is close to ideal in ensuring all sink do equal work in the network.