Litcius/Paper detail

Enhancing IoT-Edge Computation with Data Forwarding based Decentralized Deep Neural Networks

B H Pawan Prasad, Rachappa Jopate, Pankaj Savita, A Basi Reddy, Bharath Shankar, M S Arunkumar

202321 citationsDOI

Abstract

The incorporation of edge computing has become essential in the big data and Internet of Things age to tackle issues related to latency, real-time processing, and data scale. This paper delves into the imperative need for edge computing within the context of big data environments. Distributed deep learning methodologies are explored, encompassing a diverse array of deep learning models. Particularly, a novel paradigm named Data Forwarding based Decentralized Deep Neural Network (DF-DDNN) is introduced to achieve low latency IoT-Edge computation. In order to reduce latency in Io T ecosystems, the DF-DDNN model makes use of edge devices’ processing capabilities. By strategically distributing data processing and computation at the edge, this model seeks to overcome inherent limitations of conventional Io T architectures, with a keen focus on network, processing, and overall latencies. The research provides a thorough performance evaluation that compares the DF-DDNN model’s performance versus conventional IoT models in a number of areas. The outcomes demonstrate how much better the suggested DF-DDNN model is than traditional IoT systems. Notably, improvements are evident in terms of reduced network latency, accelerated processing speeds, and overall latency mitigation. This research underscores the critical importance of embracing edge computing in the context of big data and deep learning applications. The DF-DDNN model emerges as a promising avenue for advancing IoT-Edge computation, effectively addressing latency concerns and augmenting the performance of IoT-enabled systems.

Topics & Concepts

Computer scienceEdge computingEdge deviceLatency (audio)Distributed computingDeep learningBig dataContext (archaeology)ComputationEnhanced Data Rates for GSM EvolutionData processingArtificial neural networkInternet of ThingsData modelingArtificial intelligenceData scienceCloud computingEmbedded systemData miningTelecommunicationsDatabaseAlgorithmPaleontologyBiologyOperating systemIoT and Edge/Fog ComputingAdvanced Neural Network ApplicationsContext-Aware Activity Recognition Systems