Ambient self-powered cluster-based wireless sensor networks for industry 4.0 applications

Abstract

Smart grid is one of the major prospective candidates in the Industrial Internet of Things family that ensures smooth and efficient power distribution, restoration in times of emergency, and usage control for the consumers. Electric power generators contribute at the core of smart grid along with the transmission lines and transformers. Extensive research works are conducted to optimize different parameters such as efficient energy usage, automated demand response, and emergency grid failure recovery. However, the component status analysis of the electric generators within a smart grid is still in the nascent stage. In this paper, we propose a novel routing protocol for supervised device-data transfer from smart grid generators to the command and control center using wireless ad hoc and sensor networks. Our protocol assumes various sensor devices (temperature sensors, oil level sensor, turbine status sensors, etc.) to be employed on each generator due to their mechanical sophistication. Additionally, we introduce the ambient energy harvesting for the sensors energy replenishment to accommodate tolerable node outage. Our simulation results demonstrate promising outcome with respect to different key parameters such as message flow, energy consumption, outage frequency, remaining energy, and harvested energy.

Appendices

A Theoretical derivations of Basic LEADER (heterogeneous aggregation)

This section explains the theoretical derivation of Basic LEADER protocol for heterogeneous data aggregation. It is to be noted that the Basic LEADER elects only one route consisting of active nodes from a cluster.

1.1 A.1 Message flow

Let us assume that the selected route has n nodes considering no outage. Thus, the number of packets transmitted from leader (source) node to the cluster head = $\frac{n(n+1)}{2}$. It is to be noted that the length of the route may vary over the simulation time due to the node outage.

If the length of each packet is L bits, then the total message flow, $MF_r$, over a route from source through relays and to the destination is defined using Eq. 29.

$$\begin{aligned} MF_r = \frac{n(n+1)r}{2} \times L \end{aligned}$$

(29)

The normalized message flow is defined in Megabits (1 Megabits = $10^6$ bits) in Eq. 30 where r refers to the round number, and $L = 800$ bits.

$$\begin{aligned} MF_r= & {} \frac{n(n+1)r}{2n \times 10^6} \times L \\= & {} \left[ \frac{(n+1)\times L \times r}{2 \times 10^6}\right] \\= & {} \frac{(n+1)\times r \times 800}{2\times 10^6} \\= & {} (n+1) \times r \times 4 \times 10^{-4} \end{aligned}$$

$$\begin{aligned} \boxed {MF_r = (n+1) \times r \times 4 \times 10^{-4}} \end{aligned}$$

(30)

Thus, if we want to calculate the message flow over n nodes at round 1 to 100,

$$\begin{aligned} MF_1= & {} (10+1) \times 1 \times 4 \times 10^{-4} \\= & {} 0.0044 \\ MF_2= & {} (10+1) \times 2 \times 4 \times 10^{-4} \\= & {} 0.0088 \\&\ldots \\ MF_{100}= & {} (10+1) \times 100 \times 4 \times 10^{-4} \\= & {} 0.44 \end{aligned}$$

1.2 A.2 Energy consumption

Energy consumption, $EC_r$ at round r is defined in Eq. 31. Both transmit energy and reception energy is equal as transmit current, $I_t$ and reception current, $I_r$ are equal, $E_t = E_r = V \times I_t = V \times I_r = 3.3 \times 25 \times 10^{-3} = 0.00825$ Joule.

$$\begin{aligned} EC_r= & {} \left[ \frac{n(n+1)}{2} \times T_p \times (E_t+E_r) - (T_p \times E_r)\right] \times r \\= & {} \left[ \frac{n(n+1)}{2} \times T_p \times 2E_t - (T_p \times E_t)\right] \times r \\= & {} \left[ T_p \times E_t \times \{n(n+1)-1\}\right] \times r \\= & {} (n^2+n-1) \times T_p \times E_t \times r \end{aligned}$$

$$\begin{aligned} \boxed {EC_r = (n^2+n-1) \times T_p \times E_t \times r} \end{aligned}$$

(31)

We can calculate the energy consumption, EC over n nodes at round 1 to 100 considering $L=800$ bits and $R=220$ kbps.

$$\begin{aligned} EC_1= & {} (100+10-1) \times \frac{L}{R} \times 0.00825 \times 1 \\= & {} 0.0327\\ EC_2= & {} (100+10-1) \times \frac{L}{R} \times 0.00825 \times 2 \\= & {} 0.0654\\&\dots \\ EC_{100}= & {} EC_1 \times 100 \\= & {} 3.27 \end{aligned}$$

1.3 A.3 Remaining energy

$$\begin{aligned} \begin{aligned} RE_r&= E_{total} - EC_r - E_{idle} + Recharge_{vib}\\&= E_{total} - EC_r - \underbrace{(n \times E_{idle} \times T_{idle})}_\text {Idle Discharging}\\&\quad + \overbrace{(n \times E_{vib} \times \varDelta )}^\text {Active Recharging} \\ \end{aligned} \end{aligned}$$

(32)

where, discharging during the idle time, $\boxed {EC_{idle} \gg Recharge_{vib}}$

• When number of $\hbox {nodes} = n$, maximum energy capacity of $\hbox {sensors} = E_{max}$, then the total remaining energy for a particular timestamp. $E_{total} = n \times E_{max}$

• $EC_r = (n^2+n-1) \times T_p \times E_t \times r$

• $\varDelta$ = Data generation interval = 0.03

• $E_{vib}$ = Vibration energy = 0.0035

• $E_{idle}$ = Energy consumption during idle period = $2.31 \times 10^{-5}$

• $T_{idle} = \varDelta - \big [\frac{n(n+1)}{2} \times T_p \big ]= 0.3 - 0.2 = 0.1$

Thus, the cumulative remaining energy in the simulation after round number 1, $RE_1$,

$$\begin{aligned} RE_1= & {} 20-0.0327-(10 \times E_{idle} \times T_{idle}) \\&\quad + (10 \times E_{vib} \times \varDelta ) \\= & {} 20-0.327-(2 \times 10^{-5}) + 0.0105 \\= & {} \underbrace{19.9778}_\text {The actual simulation energy at round 1 is 19.9770} \end{aligned}$$

The difference between theoretical and simulation $\hbox {energy} = 19.9778 - 19.9770 = 0.0008$, this is due to the charging capacity of the sensors. For example, a node has remaining energy closest to $E_{max}$ and when it gets recharged for $\varDelta$ period, the total energy exceeds $E_{max}$. But the sensors battery will not receive charge more than $E_{max}$. For this reason, sometimes the theoretical remaining energy will show a bit more than the actual simulation results. While analyzing the theoretical results, we should consider individual node recharge and remaining energy to exactly match with the simulation results. The remaining energy depends mainly on the consumed energy, as the number of nodes and idle time are fixed before any node failure.

1.4 A.4 Outage frequency

The outage frequency refers to simply keeping track of the nodes those go below the threshold energy or have some mechanical fault. Instantaneous node outage are calculated after each $\varDelta$ time and kept in an array. The outage frequency at time t, $OF_t$ can be explained in Eq. 33 where $O_j$ refers to the total node outage at time $t+\varDelta$.

$$\begin{aligned} \underbrace{OF_{t+\varDelta }}_{t=0 \rightarrow T_{max}} = \sum _{j=1}^{n} O_j \end{aligned}$$

(33)

B Theoretical derivations of improved LEADER (heterogeneous aggregation)

As improved LEADER provides 200% increase in efficiency due to splitting of active routes into two, the derivations can be done from the basic LEADER equations. In this section, only equations are provided without the detail explanation. The parameters are similar as used in “Appendix A”.

1.1 B.1 Message flow

$$\begin{aligned} \begin{aligned} MF_r&= \frac{n(n+1)r}{4n \times 10^6} \times L \\&= \big [\frac{(n+1)\times L \times r}{4 \times 10^6}\big ] \\&= \frac{(n+1)\times r \times 800}{4\times 10^6} \\&= (n+1) \times r \times 2 \times 10^{-4} \end{aligned} \end{aligned}$$

(34)

1.2 B.2 Energy consumption

$$\begin{aligned} \begin{aligned} EC_r&= \left[ \frac{n(n+1)}{4} \times T_p \times (E_t+E_r) - (T_p \times E_r)\right] \times r \\&= \left[ \frac{n(n+1)}{4} \times T_p \times 2E_t - (T_p \times E_t)\right] \times r \\&= \left[ T_p \times E_t \times \left\{ \frac{n(n+1)}{2}-1\right\} \right] \times r \\&= \frac{1}{2} \times (n^2+n-2) \times T_p \times E_t \times r \end{aligned} \end{aligned}$$

(35)

1.3 B.3 Remaining energy and outage frequency

This is similar to Eq. 32, except the $EC_r$ value should come from Eq. 35.

$$\begin{aligned} \begin{aligned} RE_r&= E_{total} - EC_r - E_{idle} + Recharge_{vib}\\&= E_{total} - EC_r - \underbrace{(n \times E_{idle} \times T_{idle})}_\text {Idle Discharging} \\&+ \overbrace{(n \times E_{vib} \times \varDelta )}^\text {Active Recharging} \\ \end{aligned} \end{aligned}$$

(36)

Please refer to Eq. 33 for outage frequency.