Dynamic Edge Fabric EnvironmenT: Seamless and Automatic Switching among Resources at the Edge of IoT Network and Cloud

When wearable technology meets computing in future networks

“…• The exploration of the optimal displacement of network functions and network elements of the future infrastructure and remote computing. [20] Cognitive IoT gateway equipped with skills of intelligent distribution between the Edge and cloud using machine learning [19] Platform for seamless software mobility among the nodes of an Edge-cloud environment [7] Framework for scheduling applications over hybrid / heterogeneous networks Migration of services / Digital Twins A, DP, N [47] Algorithm for latency aware replica placement [60] Algorithm for optimal service migration strategy based on dynamic programming [49] Algorithm for users' workload distribution in response to movement around the MEC network based on the regularization technique [50] Real-time service migration solution based on Markov Decision Process Energy-efficient data gathering network A, N, SW [13] Adaptive compressive sensing scheme that offers simultaneous compression and encryption in a lightweight fashion [56] Data-driven compressive sensing framework for the energy-efficient wearable sensing [54] Adaptive compressed classification architecture for activity recognition Lack of network resources A, N, DP [15] D2D technology as a solution to increase system throughput by offloading data and reusing benefits [17] D2D communications and MEC system symbiosis to increase the processing power of the entire system Low quality-of-service indicator A, N [29] Cluster-based multicast methods for D2D communications Insufficient computing capabilities A, N, HW [48] Computation offloading scheme, which leverages computing resources through D2D links to improve MCC performance Discovery of inefficient computing resources A, N [48] Carefully designed access restrictions to allow users to maximize the computing resources of nearby mobile devices without spending much power on discovering other devices N, DP [2,14] Approximate and beyond approximate computing techniques Classification problems, anomaly detection, forecasting problems DP, SW [24,39,58] Applying machine learning approaches to decrease the impact on the overall execution Aspects of AI/ML on Edge A, N [4] Cross-platform framework that ensures the superior Edge AI ability N [55] ML-based authentication in IoT systems [11] Framework based on the convergence of Blockchain and AI Lack of power and computational efficiency for ML/AI eneblers SW, HW…”

“…In practice, those are sometimes viewed as mutually exclusive approaches to network infrastructure while they may function in different ways, using one does not preclude the use of others. In this context, some approaches have been proposed to address this problem (e.g., IoT cognitive gateways [20], a platform that automatically determines the best environment for executing a task [19], and a framework for scheduling applications in hybrid public-private cloud [7]). However, automatic switching between computing paradigms is, in any case, a problem for future research.…”

Distributed Edge Computing Offloading Algorithm Based on Deep Reinforcement Learning

“…In addition to the above four elements, there is a hyperparameter γ . γ is the future reward weight, the value range is [0,1]. The value function is focused on the currently obtained reward when γ tends to 0.…”

“…IN order to meet the access network requirements of different devices in different scenarios [1], network operators use heterogeneous access methods (wired or wireless access point) when they build networks. Limited by the computing capability of the IoT device itself, the IoT device will offload some tasks to the cloud server to reduce the device's own burden.…”

From Cloud to Edge: Seamless Software Migration at the Era of the Web of Things

“…On the one side, several large-scale IoT applications operate in dynamic environments: consequently, software solutions are requested to adapt to rapid changes in the bandwidth/computational resources, in the number of connected devices, and in service requirements. Several IoT platforms like [2] [3] provide such a layer of adaptation by supporting the seamless software mobility among the nodes of an edge-cloud continuum. On the other side, mobile…”