Efficient Energy Resource Selection in Home Area Sensor Networks using Non Swarm Intelligence Based Discrete Venus Flytrap Search Optimization Algorithm

A new non-swarm intelligence algorithm called discrete Venus fly-trap search algorithm (DVFS) is proposed for the optimal energy resource selection for sensor nodes in home area sensor network (HASN). DVFS algorithm is a population-based, non-swarm intelligence search algorithm that copycats the foraging behaviors of Venus fly-trap plant. The performance of DVFS based energy resource selection methodology is studied by simulating in wireless sensor network toolbox in Matlab2016. The simulation results exposed that the proposed approach can identify optimal energy resource selection from the energy source station to provide the power supply to the nodes in HASN for the network lifespan increment.


Introduction
For the past two decades, many meta-heuristic optimization algorithms are proposed to solve problems in different fields. Nowadays, these algorithms are quite popular methods because of their good computation power and easy conversion to the real problem. Metaheuristic methods are very general which can be adapted easily to any type of problems with a single objective or multi-objective.
A recent study indicates that plants also exhibit intelligent behaviors, which can be modeled mathematically for the system model and objective function. For example, foraging behavior of Venus flytrap, a carnivorous plant can be modeled as an optimal search algorithm for identifying the optimal solution under specified constraints.
Swarm Intelligence based Search algorithms inspired by social insects, fish, bird flocking, honey bees, etc., mimic direct or indirect communication among individuals, especially 1 3 information regarding promising search space for their foraging. But, many species search for food independently and autonomously rather than cooperatively. Such species also have food search stratagems to maintain the species, which is known as non-swarm intelligence.
Discrete Venus fly-trap search algorithm (DVFS) is modeled based on the foraging behavior of Venus fly-trap plant. The botanical name of Venus flytrap is Dionaea muscipula is shown in following Fig. 1. The great scientist Darwin quoted this plant as "one of the most wonderful in the world". This algorithm is devised on the rapid closure action of its traps (also called as leaves). This trap closure is due to the stimulation of the trigger hairs that present in the two lobes of the leaf by the movements of prey (small insects, small animals, raindrop, fast blowing wind, etc.). This traps in this plant search for their food independently and autonomously without any information exchange among them [1].
Energy resource selection (ERS) is techniques that can find an optimal energy source selection for a home appliance through wireless power transfer to calculate the overall network energy consumption. In this paper, discrete Venus fly-trap search algorithm (DVFS) is the new optimal search algorithm devised the first time for selecting energy sources [2]. The remaining part of the paper is organized as follows. Section 2 presents the state-of the-art of energy efficient sensor network models using optimization algorithms. Section 3 describes

Review Literature
This section describes the detailed assessment of the state-of-the-art in energy-efficient model techniques for sensor network and the issues in the existing research work related to node energy saving. Table 1 lists the some of the interesting works related to the energy saving aspects for smart home area network. Currently, the varieties of sensor network application use tiny sensor motes, which is limited battery capacity. This is one of the most important energy constraints for the increase of network lifespan. Energy Harvesting is an alternative way for giving external energy resources like solar, wind energy, vibrational, private sector, etc.
The main drawback of these energy harvesting solutions is high uncertainty, which shows frequent irregular behavior (time varies the amount of solar radiation, weather condition.). In order to overcome this, recent technology develops the wireless energy transfer in perspective of more consistent and deterministic energy resource in wireless sensor network which can also be adopted for HASN. Recent trends also focus charging the wireless sensor motes as per benchmark scheduling methodology in network time.
This research work proposes a novel on-demand wireless charging scheme for HASN using DVFS algorithm, which finds the optimal power using on-demand energy-efficient strategy. Following challenges are, discovered and feasible solutions are discussed using proposed methodology.
• To avoid node failure by maintaining the service queue buffer to store the charging the request received. • To find optimal ESS using on-demand wireless charging scheme using DVFS algorithm for HASN. • Scalability of node charging request is possible by using service queue buffer.
Moreover, 70-80% of sensor energy utilized for network communication processes in the HASN. When the sensor runs out for their residual energy, external energy source plays a vital role to supply energy to the sensor node. Hence, this work proposes the optimal energy source selection for HASN.

HASN Architecture
A home area sensor network (HASN) is a network of numerous smart home devices. These smart home devices are monitored and controlled by in-house control called utility gateway (UG) for network communication. Mainly, it is used to gather sensor information from a variety of home devices/appliances and deliver control information to these devices for efficient consumption of energy through utility gateway. The HASN has the option to turn home devices off and on conveniently for energy efficiency (e.g. automatic air conditioners on/off). Typically, HASN covers areas up to 1000 square feet and to support from 1 to 1000 kbps.  Figure 2 depicts the HASN architecture taken for the study.
Smart meter (SM) is the main controller with a wireless transceiver. It can synthesize and analyze the information received from all intelligent motes (IMs); send back data to distributed IMs, information it sends a response to the user terminal and the control signal to the utility gateway. Each node or home appliance is connected with an IM. The IM collects information from its appliance and forwards it to the smart meter.
Intelligent mote (IM) integrated into the utility gateway, wireless transmitters and attached with a variety of home sensors/devices. They are used to collect environment information, transmit sensing and control data.
Each IM can make an automatic response in consideration of the energy characteristics of its connected home device. That is it sends the communication energy request (Ereq) to the smart meter via utility gateway for active node participation in HASN.

Network Model
Suppose there area N nodes with intelligent mote, m gateway and smart meter are connected to the k energy source. Eo is the initial energy of nodes which is randomly distributed in the area on A. Figure 3 shows the first order energy model for HSAN. Equations are used to calculate the transmission and receiving energy for a 'k' bit message.
Nodes in HASN satisfy the following conditions: a. The initial energy of all nodes is the same for the network area A. b. Nodes are equipped with un-replace-able or un-rechargeable batteries. c. Each node works independently & autonomously.

Energy Resource Selection (ERS) in Energy Source Station (ESS)
Given Utility network G = (N,E) with communication data X = { X 1 ,X 2 ,…, X n ), reserved energy level Residual energy.
a. Select the energy source with maximum energy balance for communication b. Find energy source on-demand by Intelligent Mote which intimate through smart meter maximize the node lifespan

Optimal Energy Resource Selection (OERS) Strategy
The goal of OERS strategy is to maximize the fitness function in order to achieve the maximum lifetime of the nodes in the HASN by extending the battery power of the node via external energy source in ESS. where, where,

Discrete Venus Fly-Trap Search (DVFS) Algorithm
In DVFS, the rapid closure behavior of the Venus Flytrap leaves (trap) to capture the prey is mimicked. The number of traps is decided based on the problem objective. Each trap represents one energy resource. It is represented as binary strings of length n, where n is the number of resource nodes in HASN.
The presence of each node in a particular trap (Energy Resource) is depicted using this binary string. The trap parameters like trigger time (t), action potential (u t ), the charge accumulated (C), flytrap status (δ(f(t))), object status (s(C)) are initialized at the beginning of the algorithm. The trapping process is performed until the optimal solution is attained.
Initially, the trap is kept open, seeking prey. The Energy resource nodes are randomly chosen initially. When the prey has arrived, the trigger hair is stimulated first at t = 0, the second stimulation is performed at t < T (T = 30 s) then the flytrap parameter values are updated.
The fitness of the flytrap is defined as the energy balance cost of energy resource calculated using Eq. 4.

where ES i -ith energy source,Renergy(,) -Initial energy of the ES i , Energy Expend Cost (EBC)-Total Energy expenditure incur in ES i .
The fitness function is to maximize the node lifetime by selecting the energy resource with maximum energy balance. At each iteration of the EBC is evaluated; the maximum energy balance source is selected. So the fitness of the energy resource node (prey) is necessary to decide the object (prey) capture. The flytrap will be sealed. If the trap is closed as well as the object is trapped otherwise the flytrap will be reopened for the next capture. The sealed trap will not be undergone to next iteration until better flytrap than the sealed trap has arrived. The process of capturing the prey will be performed until the maximum snaps of the flytraps. After max_snap iterations, the best flytrap is returned which is the required optimal energy source for communication.

Parameter Setup
In this plant, the action potential [15] essential for the closure of the leaf. The potential generated is dissipated at a specific rate and reaches to zero. It's jumped to 0.15 V at 0.001 s and simultaneously dissipated to zero after 0.003 s. The action potential u t required for how much the energy takes and it is defined by the exponential function is given in Eq. 6 (C).
where, t-the trigger time at which the trigger hairs are stimulated, t<0 -represent before the first stimulation, t=0 -intimate for the first stimulation, t>0-means the second stimulation.
The two successive stimulations times may be less than 30 s in demand to start the flytraps are going to the closure stage. There is no action potential available for before first stimulation of the flytrap. The charge accumulation [15] can process to the stepwise growth of a bio-active substance, subsequent channel activation by the action potential. The charge accumulation will be described by a linear dynamic system is given in the Eq. 7(δ(ft)).
where is C the charge accumulated by the lobes for trap shutter, k c is network channel rate of dissipation of charge between the first two stimulations and C reaches to zero after 30 s, k a is channel rate of charge accumulation. So, it's implicit the second stimulation will be occurred within 30 s to attain the maximum charge of 14µC to shut the flytrap. In these computational of the flytraps actions, the charge accumulation is calculated based on the fitness function f (ES) given in Eq. (2). The next parameter is evaluating the status of the flytrap, used to know the present flytrap status of the plant. The flytrap status may be either 0 or 1 or 2 (open or close or seal.) is estimated using the equation u t .
Initially, the flytraps in the opened state (0). If the prey (object) triggering, the trap is closure then the trap will be in a closed state (1). The first stimulation of the time point zero is taken, the second stimulation time is representing t. The time threshold T takes two stimulations periods (say 30 s) [16]. All the closed stage of the traps won't be sealed. Only the best fitness of the flytrap will be located in the sealed stages. The reopening of the sealed flytraps takes when the fittest flytrap will be achieved.

Energy Source Station (ESS)
Energy source station is plenty of energy sources available in a station or certain area, and it's connected with smart meter (SM), used for charging the energy consumed nodes/sensors/end devices/home appliance devices. Many researchers have done a lot of research works to find two types of charging methods and they are.
Due to the uncertainty of network environmental changes, the periodical charging scheme is not suitable for the complex network environment change. Generally, the ondemand charging scheme is appropriate for the complex and changeable environments, in which, and the nodes send charging requests to the SM when the residual energy is more minimum than the predefined energy threshold (N e ), and the SM will charge the sensors/  [17]. SM maintains the service queue buffer to store all the energy charging requests and will serve the request according to the adopted on-demand charging strategy (AOCS). This AOCS ranks the sensor/node with energy requests based on their amount of the residual     . 6 Throughput of HASN energy. The priority is given to the sensor/node, which has low residual energy. At this point, to serve a request, an optimal energy source is selected for sensors/node and fills its energy supply via WPT. SM may be used to charge the power of multi nodes in the network simultaneously, based on multiple charging requests received from the charging coverage area [18]. To avoid the battery depletion in sensors/nodes a more flexible and high efficient method called energy source selection using DVFS is proposed in this work to serve the charge request of the sensors/nodes. On the whole, it prevents node failures and increases the network lifespan.

Results and Discussion
The HASN simulation done by using network simulator tool (NS 2.35). It's a tool control language and easy to generate different type of network area environment, network size, packet size/length, can create single/multiple/grid-based nodes, various topologies, interfaces, TCP/UDP to sink, CBR & FTP connection. The available protocols are AODV, AOMDV, DSDV, DSR, TORA, and one can create own protocol is also possible in ns2 tool. Table 2 shows the parameter setup for the home area sensor network taken for study in this work. Table 3 displays the performance of the proposed DVFS algorithm for Energy Source Selection (ESS) problem. Figure 4 displays the energy source station (ESS) charging the HASN nodes and variant of overall charging, individual node charging via SM decision-making approach for the energy source selection.
The performance of this work is analyzed using the relevant QoS parameter such as Packet Delivery Ratio, Throughput, Energy Consumption, Packet loss ratio. The number of nodes in the HASN is varied from 10 to 60 in the increment of 10.
PDR percentage is higher for ESS with DVFS than the normal problem. This is because of when a sensor/node with low residual energy is charged with an external energy source through WPT. Therefore, the availability of a sensor/node is increased for communication in which turn increase the PDR and throughput. Packet loss ratio of the proposed work is lesser which means that almost all the nodes in the HASN are the inactive state for the network communication.
The energy consumption of the proposed work is a little bit higher which is evident from Table 3. This is due to less number of nodes are in the inactive state. The nodes in the in-active state do not participate in the communication. Due to this, the standard algorithm has less energy consumption than ESS with DVFS algorithm. Figure 5 shows the packet delivery ratio of home area sensor networks with/without applying the Discrete Venus Fly-Trap Search Algorithm in the given NS-2 simulation environment. Figure 5 presents the Throughput of the different home appliances (nodes) in home area sensor networks.
Tables 4, 5 and 6 show the performance of the proposed work in terms of throughput (in Kbps), energy consumption (in Joules) and packet loss ratio (in Packets/µ sec). From the above tables, it has been observed that the proposed DVFS algorithm has the capability to identify the optimal energy source for the single/multi sensor(s)/node(s) in the HASN efficiently in order to increase network lifespan. Figures 6, 7, 8 are the graphical representations of Tables 4, 5, 6 in different home appliances using energy-consumption and packet loss ratio of home area sensor networks with/ without applies the discrete venus fly-trap search algorithm.

Conclusion
There are many successful searches, optimization algorithms and techniques in the literature for energy-efficient sensor network model. But still, design, development, and implementation of new techniques are a key for improvement in the scientific. Even the benchmark algorithm cannot give the best results for all of the problems. Moreover, all existing search or optimization algorithm cannot give the best result for all real-world problems of a different nature. That is reasons new DVFS algorithms have been proposed for energy source selection for HASN. The results showed that the optimal selection of the energy source has been done to provide the power supply to the nodes in HASN for the increased network lifespan.