Energy efficient architecture for mitigating the hot-spot problem in wireless sensor networks

Abstract

In wireless sensor networks (WSNs), mutual coordination of cluster heads (CHs) is essential to transmit their data towards the sink node through many-hop fashion. As a result of this activity, the CHs in close vicinity to the BS are overburdened with massive relay traffic, which establishes a hot-spot problem. In this paper, in order to capture the hot-spot problem, harris hawk optimization (HHO) based algorithms have been proposed, jointly termed as HHO-UCRA (HHO build on unequal clustering and routing algorithms). In the first step, CH selection mechanism has been proposed based on HHO based technique. Afterwards, the derived CH_Assignment function is used for the cluster formation. Finally, efficient hawk encoding schemes and novel fitness functions of HHO based technique have been formulated for both the algorithms. In the extensive simulation, HHO-UCRA is executed with varying number of sensors and CHs for all the WSN scenarios. Thereafter, the proposed algorithm is evaluated with some recent existing routing approaches and standard meta-heuristic based approach known as PSO-UCRA, to show the efficiency in terms of benchmark indicators of WSNs, such as network energy consumption, lifetime of network, convergence rate, data packets received by the BS and the number of alive nodes.

Appendix 1: Harris hawks optimization (HHO)

In the swarm intelligence optimization algorithm group, harris hawks optimization is one of the recent innovation, which is developed by Heidari et al. It is framed using the hunting strategy of hawks birds. Harris hawks shows the smart behavior while attacking the prey. They starts from monitoring the prey and ends after successfully attacking the prey. In this process, each hawk knows the position of other group members by the help of which hunting process can be executed.

HHO is composed of two phases. First one is diversification and another is intensification. Both the phases together mimics the hunting strategy of prey. First phase also consists of two sub-phases: (1) diversification and (2) process of switching from diversification to intensification. Intensification phase is based on various strategy, namely, (1) soft besiege, (2) hard besiege, (3) soft besiege with progressive quick pounce and (4) hard besiege with progressive quick pounce. Among all these strategies, one of them is taken into the consideration dynamically according to the situation. The detail description of both the phases are as follows.

1.1 1.1: Diversification

In this process, basic nature of harris hawks (HH) is considered, they keep an eye on desert site to track and detect a prey. In order to detect the prey, harris hawks perch on random locations, it is found on two strategies. In the first strategy, perch depends on the other family members and position of rabbit, which is shown in Eq. 29 when $d<0.5$. In the second strategy, harris hawks are sitting in random tall trees for the perching, which is shown in Eq. 29 when $d\ge 0.5$.

$$\partial (t + 1) = \left\{ {\begin{array}{*{20}l} {{\mathbf{if}}~d \ge 0.5~{\mathbf{then}}} \\ {\partial _{{rand}} (t) - r_{1} (\partial _{{rand}} (t) - 2r_{2} \partial (t))} \\ {{\mathbf{Otherwise}}...} \\ {(\partial _{{rabbit}} (t) - \partial _{m} (t)) - r_{3} (LB + r_{4} (UB - LB))} \\ \end{array} } \right.$$

(29)

$$\begin{aligned}&\partial _m(t)=\frac{1}{N} \sum _{i=1}^{N}\partial _i(t) \end{aligned}$$

(30)

where $\partial (t+1)$ represents the position of hawks in the next generation, $\partial _{rabbit}$ is position of rabbit, $\partial (t)$ denotes the current generation vector, various random numbers generated between 0 and 1 is denoted by $r_1$, $r_2$, $r_3$, $r_4$ and d, randomly chosen hawk is represented by $\partial _{rand}$, and upper and lower bound of variable is shown by LB and UB. The average locations of hawks is represented by $\partial _m(t)$.

1.2 1.2: Switching from diversification to intensification

Initially, optimization techniques explore the search space i.e., diversification. Thereafter, exploits the neighborhood of the solution (intensification). This phenomena is modeled in HHO using the escaping energy (E) of prey (rabbit). If $|E|\ge 1$ then diversification is performed, when $|E|<1$ intensification is taken into the consideration. Hence, switch from diversification to intensification is performed using the escaping energy or prey. The estimation of energy is shown in Eq. 31.

$$\begin{aligned} E=3E_o(1-\frac{t}{T}) \end{aligned}$$

(31)

Fig. 9

Demonstration of diversification and intensification process

1.3 Intensification

This process is composed of four strategies, namely, (a) soft besiege, (b) hard besiege, (c) soft besiege with progressive quick pounce and (d) hard besiege with progressive quick bounce. The detail description of all these strategies are as follows.

(a) Soft Besiege: In the soft besiege, prey have enough energy to escape, but, failure to escape. This phenomena is modeled using the Eq. 32, when $r\ge 0.5$ and $|E| \ge 0.5$.

$$\begin{aligned}&\partial (t+1)=\delta \partial (t)- E(J \times \partial _{rabbit}(t)-\alpha (t)) \end{aligned}$$

(32)

$$\begin{aligned}&\delta \partial (t)= \partial _{rabbit}(t)-\partial (t) \end{aligned}$$

(33)

$$\begin{aligned} J=2 \times (1-r_5)and \end{aligned}$$

(34)

where $r_5$ is a random number between 0 and 1, and J represents the jump strength.

(b) Hard Besiege: In the hard besiege, prey/rabbit is not having enough energy to escape and hawk performs the surprise pounce on it. This scenario is modeled using the Eq. 35, when $r\ge 0.5$ and $|E| < 0.5$.

$$\begin{aligned} \partial (t+1)= \partial _{rabbit}(t)- E (\delta \partial (t)) \end{aligned}$$

(35)

(c) soft Besiege Strategy with Progressive Quick Pounce: In this strategy, prey has sufficient amount of energy to escape and still soft besiege is modeled over it. This case is more intelligent than the aforementioned case, when $|E|\ge$ and $r< 0.5$. This scenario is modeled using the levy flight function, which is shown in Eq. 36.

$$\begin{aligned} \partial (t+1)= \left\{ \begin{matrix} Y &{} if ~F(Y)< F(\partial (t)) \\ Z &{} F(Z) < F(\partial (t)) \end{matrix}\right. \end{aligned}$$

(36)

Here, Y decides the hawks next movement, which can be estimated as follows.

$$\begin{aligned} Y=\partial _{rabbit}(t)- E(J \times \partial _{rabbit}(t)- \partial (t)) \end{aligned}$$

(37)

When hawk approaching towards the prey, then prey and hawk both sometimes perform random movement. This pattern is modeled using LF function as follows.

$$\begin{aligned} Z= Y + S \times LF (D) \end{aligned}$$

(38)

$$\begin{aligned} LF(x)= 0.01\times \frac{u\times \gamma }{{|v|}^{\frac{1}{\omega }}}, \gamma = \left( {\begin{array}{c}\Gamma (1+\omega )\times sin(\frac{\pi \omega }{2}) \\ \Gamma \frac{1+\omega }{2}\times \omega \times 2^{\frac{\omega -1}{2}}\end{array}}\right) ^{\frac{1}{\omega }} \end{aligned}$$

(39)

where u, v are random number between 0 and 1, and $\omega$ is a constant value.

(d) Hard Besiege: Strategy with Progressive Quick Pounce In this strategy, prey not having sufficient energy to escape and hard besiege is modeled over it, when $|E|<0.5$ and $r<0.5$. This scenario is represented using the Eq. 40.

$$\begin{aligned} \partial (t+1)= \left\{ \begin{matrix} Y &{} if ~F(Y)< F(\partial (t)) \\ Z &{} F(Z) < F(\partial (t)) \end{matrix}\right. \end{aligned}$$

(40)

The estimation of Y and Z values are as follows.

$$\begin{aligned}&Y=\partial _{rabbit}(t)- E(J \times \partial _{rabbit}(t)- \partial _m(t)) \end{aligned}$$

(41)

$$\begin{aligned}&Z= Y + S \times LF (D) \end{aligned}$$

(42)

Illustration of figure 9: The preliminary phase of harris hawk optimization is Identification (to generate random solutions i.e r1, r2, r3, r4) which have been shown in Fig. 9a. In this phase, the hawk identifies its target and its vicinity. The next step can have two outcomes to proceed which depends upon the energy of prey $E_r$. If $E_r$ is greater than equal to one, then firstly diversification phase (exploration) takes place where the hawks exhausts the prey leaving with very low energy is shown in Fig. 9b, followed by intensification (exploitation) phase which is shown in Fig. 9c. But, if initially the energy of prey $E_r$ is less than one, then directly intensification takes place. In the Intensification phase there are four possibilities of strategies which take place namely, soft besiege strategy (SBS), hard besiege strategy (HBS), soft besiege and progressive quick pounce strategy and hard besiege and progressive quick pounce strategy which are shown in Fig. 9d–g respectively. All of these strategies have been discussed briefly in Sect. 2. The final outcome depends upon the chance of prey to escape (r). If r is smaller than 0.5 then the prey escapes successfully and if the value of r is greater than or equal to 0.5 then the prey cannot escape.