Clonal selection algorithm for energy minimization in software defined networks

Highlights

• Developing a multi-objective optimization problem for software defined network.

• Applying Clonal Selection Algorithm (CSA) for solving the said optimization problem.

• Validating the proposed solution with standard benchmark functions.

Abstract

With the advancements of Information and Communication Technologies (ICT), large scale distributed computing and massive data center infrastructures are becoming more common these days. Such trends have drastically put a lot of load on the volumes of data transferred over networks, thus necessitating close to capacity link utilizations flexible forwarding decision-making. Software-defined networking (SDN), with its inherent segregation of control and data planes, provides flexible decision making that can leverage the global network information available to the SDN controller for dynamic and accurate solutions. However, contemporary researchers have focused on the flexibility and security aspects of SDN, widely ignoring energy consumption strategies in next-generation IP networks, which otherwise is a crucial driver in any research field. The scanty existing energy minimization strategies are mostly based on aggregate traffic, which leads to imbalanced utilization of links and affects the quality of service (QoS) adversely. In this paper, we leverage SDN’s key benefits for reducing energy network consumption while realizing dynamic load balance with a few QoS constraints. To this end, a multiobjective optimization problem (MOOP) is formulated that attempts to minimize power consumption and link utilization. With different capacities of switches and links, finding optimal configurations and deciding best paths, even for relatively small networks, become computationally challenging and is, in fact, an NP-hard problem. In this paper, we propose to employ the Clonal Selection Algorithm (CSA), a discrete, metaheuristic solution to find out optimal solutions for this MOOP, namely a Clonal Selection based Energy Minimization (CSEM). Simulations have been carried out for testing the efficacy of the proposed CSEM using real-life network topologies and link-traffic data. The results obtained by the proposed CSEM prove to be efficacious, and the same have also been validated with three different benchmark functions to test the suitability of SDN for CSA.

Introduction

With the ever-increasing number of connected devices over the Internet, massive volumes of data are getting generated from a myriad of applications. This has led to network requirements being very stringent, and network control policies more demanding. Traditional network devices with the juxtaposed decision and forwarding planes prove incapable of meeting the desired Quality of Service (QoS), particularly for real-time constrained applications [1], [2]. Software-defined networking (SDN) provides a novel network architecture that sets apart the control and forwarding planes [3]. The separation of the control plane from the forwarding plane makes the network programmable by configuring it based on the specific network requirements [4].

With the southbound based Application Programming Interface (API), called OpenFlow, the controller communicates with the forwarding devices as shown in Fig. 1 and thus collects huge statistics from the devices [5]. Aggregation of massive volumes of network statistics at the controller and delivering control decisions based on analytics carried out on such huge data necessitates complex solutions. This implies exploring a huge solution space across multiple objectives. The controller thus faces scalability issues while customizing the network based on the specific application requirements [6]. Evolutionary algorithms (EA) are stochastic techniques that draw inspiration from natural biological evolution and social behavior of species to provide solutions to problems involving massive solution space explorations, multiple objectives, and adaptation to any unknown environment [7]. EAs tend to find near optimum or close to optimum solutions in a diverse set of solutions. EAs have addressed several issues in SDN, such as routing, security, controller placement [8]. A critical issue that arises in a network is how to lower the cost of running a network and improvement in link utilization that is over-provisioned and redundant [9]. Networks are designed to be unnecessary, viz., having more than one path to reach from source to destination. The average utilization of links inside the network tends to be low at around 30%–50%. The lowering of operating costs involved in a network can be done by designing an energy optimization algorithm which turns off links with lesser utilization by splitting the flows onto other links. The improvement in the link utilization needs to be taken care of since splitting the flows from underutilized links to other links might degrade the QoS of an application running inside the network. With heterogeneous capacities of switches and links, finding optimal configurations and deciding best paths, even for relatively small networks, become computationally challenging and is, in fact, an NP-hard problem. This implies that no polynomial-time solutions can be obtained for such a problem. Hence, we have to try for heuristic solutions in this regard. Several researchers have focused on designing energy optimization algorithms and load balancing inside the network. Most of the works have addressed energy optimization and load balancing in the network separately [10], [11], [12]. Although a few among these have addressed load balancing and energy optimization together yet the same has remained a core issue. Earlier works have used multi-objective particle swarm optimization (PSO) while designing solutions [9], [13], [14]. However, PSO based solutions suffer from partial optimization incurred by a lack of regulation in particle speed and direction. For this particular problem, PSO is not suitable since it has to be discretized first, whereas CSA is inherently a discrete optimization method. Thus it is more suited for this problem. This paper addresses both energy-saving and load balancing in a network by formulating a multi-objective clonal selection algorithm (CSA) [15]. The key contributions of our paper are as follows:

• Formulate a multi-objective optimization problem with energy minimization and load balancing as objective functions and pertinent constraints.

• Since the said the multi-objective problem is NP-hard; hence a novel multi-objective CSA, namely Clonal Selection based Energy Minimization (CSEM) for solving optimum power-consumption and link-utilization simultaneously is introduced.

• Simulations under different scenarios and real networks’ traffic data are carried out to obtain results that are further to test the suitability through three standard benchmark functions.

Fig. 2 depicts the main contributions of this paper at the conceptual junction point of existing problems and available technologies with the shaded box. For instance, the top circle denotes the fact that optimization problems exist in the literature for the two objective functions used in this paper. However, the bottom circle implies that due to CSA’s discrete nature, it is suitable for application for the said problem. The other two blue circles depict the originality of this work vis-a-vis the other two circles.

We believe our work is the first to address energy and load balancing with the use of CSA. The proposed algorithm has shown its capability to work with multiple objectives. However, the problem was mainly targeted as an energy minimization problem satisfying QoS constraints, and the said idea has successfully been implemented through the proposed algorithm. Experiments conclude that the proposed CSA performs better with improved accuracy, and the results are validated by running several benchmark functions.

The rest of this paper is organized as follows. Section 2 presents related work from recent literature that has had motivated in this research. The proposed multi-objective optimization problem using CSA with the problem formulation has been discussed in Section 3. Section 4 discusses the solution methodology with the proposed algorithm. Section 5 presents the simulation experiments and analyzes the results obtained. Also, benchmark tests were carried out to test the validity of the proposed problem. Finally, Section 6 concludes this paper.