Edge computing is a new paradigm enabling intelligent applications for the Internet of Things (IoT) using mobile, low-cost IoT devices embedded with data analytics. Due to the resource limitations of Internet of Things devices, it is essential to use these resources optimally. Therefore, intelligence needs to be applied through an efficient deep learning model to optimize resources like memory, power, and computational ability. In addition, intelligent edge computing is essential for real-time applications requiring end-to-end delay or response time within a few seconds. We propose decentralized heterogeneous edge clusters deployed with an optimized pre-trained yolov2 model. In our model, the weights have been pruned and then split into fused layers and distributed to edge devices for processing. Later the gateway device merges the partial results from each edge device to obtain the processed output. We deploy a convolutional neural network (CNN) on resource-constraint IoT devices to make them intelligent and realistic. Evaluation was done by deploying the proposed model on five IoT edge devices and a gateway device enabled with hardware accelerator. The evaluation of our proposed model shows significant improvement in terms of communication size and inference latency. Compared to DeepThings for 5 X 5 fused layer partitioning for five devices, our proposed model reduces communication size by ~ 14.4% and inference latency by ~16%.
The energy infrastructure is one of the crucial aspect in realization of sustainable smart cities. In pursuit of achieving optimized energy supply and demand, the deployment of smart grids is becoming a strategic act in many countries. There is a void of telecommunication and information support in smart grid. Adopting Information and Communication Technology (ICT) and Internet of Things (IOT) will transform the existing grids into a smarter one. In this paper we propose a framework for smart-grid based on IOT and ICT solutions with an aim towards improving energy conservation, reducing operational cost and empowering the customer with usage analytics. A 3 tier cognitive architecture consisting of Perception layer, Attention-Memory layer and Decision layer is introduced. Various ICT and IOT solutions applicable at these 3 layers are discussed. This architecture can pave a way for future intelligent Grids there by upgrading level of urbanisation.
Biological systems have many key answers for our current limitations in scaling. Miniaturization and more speed were the driving forces for VLSI technology in past few decades. As we are reaching the dead end of Moore's law, now the paradigm has shifted towards intelligent machines. Many efforts are made to mimic the commonsense observed in animals. Automation and smart devices have taken a great momentum in recent years, but to imitate human intelligence in machines with existing hardware is not viable. By only developing complex algorithms and implementing software, intelligence is not accomplishable. This paper brings out the very basic differences between brain and computers. The complexity of human brain and flaws of current computing architecture is discussed. Neuromorphic Engineering emerges as a realistic solution whose architecture and circuit components resemble to their biological counterparts. This paper acts as a primer for Neuromorphic engineering. Brain functionality being analogous, the limitation of only digital systems is addressed by the mixed mode operation of ICs. Modelling of Neuron, various software and hardware available to realize these morphed architectures are studied. The gap between the software simulation and Hardware emulation, FPGA and VLSI implementation is debated. To reach to a larger audience the paper exposes many limitations of Neuromorphic engineering and come out with many open research issues.
The circuits inspired by organising principles of brain has been proven to be faster and more energy-efficient than modern day Boolean CMOS computations. Investigations revealed that most of our limitations in current CMOS technology is due to architectural difference. Neuromorphic Engineering is one such alternative approach, which aims to reproduce spike-based computation of brain using custom VLSI circuits (as its sub-threshold region operation resembles neural behaviour). Here we aim at developing neural circuits and system that emulate basic functions of sensory systems. We start with implementation of early sensory processing like detection and transformation of input stimuli and slowly move towards cognitive processing area with introduction of a neural network (neural oscillator) for sensory segmentation and finally to implement the neural model to exhibit learning which is fundamental in various sensory functions. The learning is governed by spike-timing dependent plasticity (STDP). The operation of the systems are investigated through theoretical analysis, numerical simulations and circuit simulations.