Load-Balanced and QoS-Aware Software-Defined Internet of Things

Abstract

Internet of Things (IoT) offers a variety of solutions to control industrial environments. The new generation of IoT consists of millions of machines generating huge traffic volumes; this challenges the network in achieving the Quality of Service (QoS) and avoiding overload. Diverse classes of applications in IoT are subject to specific QoS treatments. In addition, traffic should be distributed among IoT servers based on their available capacity. In this paper, we propose a novel framework based on Software-defined Networking (SDN) to fulfill the QoS requirements of various IoT services and to balance traffic between IoT servers simultaneously. At first, the problem is formulated as an Integer Linear Programming (ILP) model that is NP-hard. Then, a predictive and proactive heuristic mechanism based on time-series analysis and fuzzy logic is proposed. Afterwards, the proposed framework is implemented in a real testbed, which consists of the Open vSwitch, Floodlight controller, and Kaa servers. To evaluate the performance, various experiments are conducted under different scenarios. The results indicate the improved IoT QoS parameters, including throughput and delay, and illustrate the non-occurrence of overload on IoT servers in heavy traffic. Furthermore, the results show improved performance compared to similar methods.

A QoS Optimization Technique with Deep Reinforcement Learning in SDN-Based IoT

“…The configuration setting is shown in TAable 1. To evaluate the performance of the proposed algorithm, we compare simulation results with the random method, Round-robin (RR) algorithm [24] in which resource allocates in turn to the UEs and each UE cannot be served in two successive periods and the work in [5]. First, we create a topology based on the proposed architecture, as shown in Fig.…”

“…The performance of the proposed algorithm is compared with Round Robin and Random methods and the [5] method, in terms of throughput and delay. Fig.…”

“…several works have been done regarding IoT and QoS issues by using networking emerging technologies such as SDN [5]- [10]. In traditional networks, hardware devices manage connections between sources and destinations based on the static and inflexible parameters.…”

“…• Proposing a deep reinforcement learning algorithm for solving resource allocation between clients and servers in an SDN-based IoT. Our simulations and evaluations results show that the proposed solution improves QoS in terms of delay and throughput compared to Random and Round Robin methods, and also show that the performance of the proposed method is near the fuzzy and predictive method in [5]. The remainder of the paper is organized as follows: Section 2 reviews some related works.…”

Hybrid Auto-Scaled Service-Cloud-Based Predictive Workload Modeling and Analysis for Smart Campus System

“…In the study [26], the authors introduced a new softwaredefined networking (SDN) architecture to address the QoS requirements of different IoT services and handle traffic among IoT servers concurrently. The problem is defined as an integer linear programming (ILP) model, which is NP-hard.…”