FPGA Implementation of Deep Leaning Model for Video Analytics

In recent years,deep neural networks have become a fascinating and influential research subject,and they play a critical role in video processing and analytics.Since,video analytics are predominantly hardware centric,exploration of implementing the deep neural networks in the hardware needs its brighter light of research.However,the computational complexity and resource constraints of deep neural networks are increasing exponentially by time.Convolutional neural networks are one of the most popular deep learning architecture especially for image classification and video analytics.But these algorithms need an efficient implement strategy for incorporating more real time computations in terms of handling the videos in the hardware.Field programmable Gate arrays(FPGA)is thought to be more advantageous in implementing the convolutional neural networks when compared to Graphics Processing Unit(GPU)in terms of energy efficient and low computational complexity.But still,an intelligent architecture is required for implementing the CNN in FPGA for processing the videos.This paper introduces a modern high-performance,energy-efficient Bat Pruned Ensembled Convolutional networks(BPEC-CNN)for processing the video in the hardware.The system integrates the Bat Evolutionary Pruned layers for CNN and implements the new shared Distributed Filtering Structures(DFS)for handing the filter layers in CNN with pipelined data-path in FPGA.In addition,the proposed system adopts the hardware-software co-design methodology for an energy efficiency and less computational complexity.The extensive experimentations are carried out using CASIA video datasets with ARTIX-7 FPGA boards(number)and various algorithms centric parameters such as accuracy,sensitivity,specificity and architecture centric parameters such as the power,area and throughput are analyzed.These results are then compared with the existing pruned CNN architectures such as CNN-Prunner in which the proposed architecture has been shown 25%better performance than the existing architectures.

Enhanced Adaptive Approach of Video Coding at Very Low Bit Rate Using MSPIHT Algorithm

Nowadays video coding approach is a major key in many applications for easy transmission and storage consumption. The process of transformation is based on the empirical wavelet transform (EWT). The encoding process of video data provides secure and less consumption of storage and the reconstruction process consists of the reverse process with the extraction. In this paper, the coding of video is carried out at a very low bit rate with the enhancement of performance by proposing an approach of modified Set Partitioning in Hierarchical Tree (MSPIHT). This method encodes the high frequency frames with the scheduling of wavelet transform for efficient performances of encoding and improves the ability of both the frequency and time. By applying empirical wavelet transform on each video frame, the component of video frequency is extracted and the low frequency frame is encoded by the H.264/AVC standard. The low coefficient values are ignored in applying the threshold and in the reconstruction process, HBLPCE method is used for imaging enhancement. The simulation of the proposed approach analysis shows better performance in reliable process and efficiency when compared to existing.

Improved QoS-Secure Routing in MANET Using Real-Time Regional ME Feature Approximation

Mobile Ad-hoc Network(MANET)routing problems are thoroughly studied several approaches are identified in support of MANET.Improve the Quality of Service(QoS)performance of MANET is achieving higher performance.To reduce this drawback,this paper proposes a new secure routing algorithm based on real-time partial ME(Mobility,energy)approximation.The routing method RRME(Real-time Regional Mobility Energy)divides the whole network into several parts,and each node’s various characteristics like mobility and energy are randomly selected neighbors accordingly.It is done in the path discovery phase,estimated to identify and remove malicious nodes.In addition,Trusted Forwarding Factor(TFF)calculates the various nodes based on historical records and other characteristics of multiple nodes.Similarly,the calculated QoS Support Factor(QoSSF)calculating by the Data Forwarding Support(DFS),Throughput Support(TS),and Lifetime Maximization Support(LMS)to any given path.One route was found to implement the path of maximizing MANET QoS based on QoSSF value.Hence the proposed technique produces the QoS based on real-time regional ME feature approximation.The proposed simulation implementation is done by the Network Simulator version 2(NS2)tool to produce better performance than other methods.It achieved a throughput performance had 98.5%and a routing performance had 98.2%.

HELP-WSN-A Novel Adaptive Multi-Tier Hybrid Intelligent Framework for QoS Aware WSN-IoT Networks

Wireless Sensor Network is considered as the intermediate layer in the paradigm of Internet of things(IoT)and its effectiveness depends on the mode of deployment without sacrificing the performance and energy efficiency.WSN provides ubiquitous access to location,the status of different entities of the environment and data acquisition for long term IoT monitoring.Achieving the high performance of the WSN-IoT network remains to be a real challenge since the deployment of these networks in the large area consumes more power which in turn degrades the performance of the networks.So,developing the robust and QoS(quality of services)aware energy-efficient routing protocol for WSN assisted IoT devices needs its brighter light of research to enhance the network lifetime.This paper proposed a Hybrid Energy Efficient Learning Protocol(HELP).The proposed protocol leverages the multi-tier adaptive framework to minimize energy consumption.HELP works in a two-tier mechanism in which it integrates the powerful Extreme Learning Machines for clustering framework and employs the zonal based optimization technique which works on hybrid Whale-dragonfly algorithms to achieve high QoS parameters.The proposed framework uses the sub-area division algorithm to divide the network area into different zones.Extreme learning machines(ELM)which are employed in this framework categories the Zone’s Cluster Head(ZCH)based on distance and energy.After categorizing the zone’s cluster head,the optimal routing path for an energy-efficient data transfer will be selected based on the new hybrid whale-swarm algorithms.The extensive simulations were carried out using OMNET++-Python userdefined plugins by injecting the dynamic mobility models in networks to make it a more realistic environment.Furthermore,the effectiveness of the proposed HELP is examined against the existing protocols such as LEACH,M-LEACH,SEP,EACRP and SEEP and results show the proposed framework has outperformed other techniques in terms of QoS parameters such as network lifetime,energy,latency.

Spectrum Sensing Using Optimized Deep Learning Techniquesin Reconfigurable Embedded Systems

The exponential growth of Internet of Things(IoT)and 5G networks has resulted in maximum users,and the role of cognitive radio has become pivotal in handling the crowded users.In this scenario,cognitive radio techniques such as spectrum sensing,spectrum sharing and dynamic spectrum access will become essential components in Wireless IoT communication.IoT devices must learn adaptively to the environment and extract the spectrum knowledge and inferred spectrum knowledge by appropriately changing communication parameters such as modulation index,frequency bands,coding rate etc.,to accommodate the above characteristics.Implementing the above learning methods on the embedded chip leads to high latency,high power consumption and more chip area utilisation.To overcome the problems mentioned above,we present DEEP HOLE Radio sys-tems,the intelligent system enabling the spectrum knowledge extraction from the unprocessed samples by the optimized deep learning models directly from the Radio Frequency(RF)environment.DEEP HOLE Radio provides(i)an opti-mized deep learning framework with a good trade-off between latency,power and utilization.(ii)Complete Hardware-Software architecture where the SoC’s coupled with radio transceivers for maximum performance.The experimentation has been carried out using GNURADIO software interfaced with Zynq-7000 devices mounting on ESP8266 radio transceivers with inbuilt Omni direc-tional antennas.The whole spectrum of knowledge has been extracted using GNU radio.These extracted features are used to train the proposed optimized deep learning models,which run parallel on Zynq-SoC 7000,consuming less area,power,latency and less utilization area.The proposed framework has been evaluated and compared with the existing frameworks such as RFLearn,Long Term Short Memory(LSTM),Convolutional Neural Networks(CNN)and Deep Neural Networks(DNN).The outcome shows that the proposed framework has outperformed the existing framework regarding the area,power and time.More-over,the experimental results show that the proposed framework decreases the delay,power and area by 15%,20%25%concerning the existing RFlearn and other hardware constraint frameworks.

Efficient Routing Protocol Based on Security for Wireless Sensor Networks

Nowadays, the major part and most standard networks usually used in several applications are Wireless Sensor Networks (WSNs). It consists of different nodes which communicate each other for data transmission. There is no access point to control the nodes in the network. This makes the network to undergo severe attacks from both passive and active devices. Due to this attack, the network undergoes downgrade performance. To overcome these attacks, security based routing protocol is proposed with the security based wormhole detection scheme. This scheme comprises of two phases. In this approach, the detection of wormhole attacks is deployed for having correct balance between safe route and stability. Also, to ensure packets integrity cryptographic scheme is used as well as authenticity while travelling from source to destination nodes. By extensive simulation, the proposed scheme achieves enhanced performance of packet delivery ratio, end to end delay, throughput and overhead than the existing schemes.

Capacitor Pattern H-Bridge Multilevel Inverter (CPHMLI) Using Phase Diposition Pulse Width Modulation for Grid Applications

This work presents an implementation of an innovative single phase multilevel inverter using capacitors with reduced switches. The proposed Capacitor pattern H-bridge Multilevel Inverter (CPHMLI) topology consists of a proper number of Capacitor connected with switches and power sources. The advanced switching control supplied by Pulse Width Modulation (PDPWM) to attain mixed staircase switching state. The charging and discharging mode are achieved by calculating the voltage error at the load. Furthermore, to accomplish the higher voltage levels at the output with less number of semiconductors switches and simple commutation designed using CPHMLI topology. To prove the performance and effectiveness of the proposed approach, a set of experiments performed under various load conditions using MATLAB tool.