Evolution-Based Deployment Scheme for Green Internet of Things Aoun Hussain, Faraz Ahmed Khan, Akhlaque AHMAD Dhanani School of Science and Engineering, Habib University, Karachi, Pakistan Email: ah03974@st.habib.edu.pk, fk03983@st.habib.edu.pk, akhlaque.ahmad@sse.habib.edu.pk Abstract—In this generation of advance technologies, where communication between physical devices is vital, Internet of Things(IoT) play an important part in connecting the cyber and physical world. IoT is used in various applications such as agriculture, smart home and smart cities, through the de- ployment of Wireless Sensor Networks(WSNs). Although, in hostile environments, where human intervention is impossible, the life cycle of a deployed WSN becomes critical. Conserving the energy consumed by a wireless sensor network, is imperative, in prolonging the life cycle of the network. This paper addresses the challenging issue of minimizing the energy consumption of WSNs on a large scale. The contributions made by this paper are using the optimization model proposed in [25] to compare Genetic Algorithm(GA) and the Mixed Integer Linear Programming(MILP) algorithm, to solve the minimum energy consumption problem for an IoT. The MILP and GA approach in solving the minimum energy consumption problem, is flexible and efficient, and helps us to achieve our goal, i.e. minimum energy consumption and maximum network lifetime of a deployed WSN. I. I NTRODUCTION Internet of Things (IoT) is a vital part of the contemporary society, which helps us to communicate with technology in real-time, bridging the gap between internet applications and humans with tools like Artificial Intelligence (AI), actuators and cameras. The implementation of an IoT on a large- scale [1] is challenging and diverse for different deployment schemes. Since IoT consists of many objects that consume high power, power consumption of an IoT plays a vital role, in choosing the best deployment technique. Progress has been made in deploying energy-efficient WSNs, such as exact [2]- [4], ad hoc [5]-[7], hierarchy [8]-[10], and hybrid [9]-[11], but these schemes have failed to align themselves with the principles of green networking, thus resulting in a non-scalable and unsustainable IoT. Therefore, the research focus of this paper is to cost effectively deploy a green wireless sensor network, in a way that would prolong the network lifetime and consume minimum energy, with minimum pollution added to the environment.The contributions made by this paper are summarized as follows: 1) Based on the proposed framework in [25], we use the optimization model for deploying a green IoT by con- sidering maximum number of sensors per relay S max , maximum number of active relays R max and minimum energy consumption E min of the network. 2) We solve the minimum energy consumption problem through Mixed Integer Linear Programming(MILP). 3) We propose an evolution-based algorithm to solve the minimum energy consumption problem, and use MILP to set a benchmark result. 4) We compare our proposed algorithms of MILP and GA, and highlight the advantages of using an evolution-based algorithm to solve the minimum energy problem. MILP is maximizing or minimizing a linear function subject to linear constraints on the variables, where one or more variables are restricted to a set of positive integers. We have transformed the model from [25] into linear equations and constraints to solve the energy problem, whereas an evolution-based algorithm is an algorithm based on natural- selection inspired by biology and genetics [28]. The remainder of this paper is organized as follows. Section II discuses the progress that has already been made on WSNs and explains how this model [25] is better. Section III describes the system framework for placing network elements in IoT. Section IV formulates the problem of green IoT deploy- ment and formally presents the optimization model. Section V introduces the MILP algorithm to solve the minimum energy consumption problem. Section VI uses an evolution-based GA to solve the same energy problem, and gives quality results. Section VII discuses the experimental results of both, GA and MILP, introduced in sections V and VI. Section VIII concludes this paper. II. RELATED WORK In this paper, we adopt the hierarchical deployment scheme with certain changes from [25], that would cater to prolong the network lifetime and facilitate our algorithm. Recently, a lot of work has been done on energy saving with the deployment of WSNs. WSNs can be classified into five categories on the basis of energy saving techniques, i.e. updating operating system [13]-[14], controlling transmitting power [15]-[17], managing duty cycle [18], [19], routing with minimized power [20]- [21], and clustering for data aggregation [10],[22]. In order to address the new challenges of energy saving in IoT, this paper presents a model that is better than previous studies in the following three ways. First, the model has considered energy saving of a node for transmitting and receiving data. Second, it has considered the data flow as link flow traffic, and addresses the energy consumption problem by limiting the 978-1-5386-4980-0/19/$31.00 © 2019 IEEE 237