An Optimization Approach to Multi-Sensor Operation for Multi-Context Recognition

. 2021 Oct 15;21(20):6862. doi: 10.3390/s21206862

Algorithm 1 Online algorithm for sensor selection with sensing schedules.

Inputs: $x_{l}^{j}$ , $π_{l}$ , $S^{m}$ , $E^{m}$ , $E_{G}^{j}$ , $D_{G}^{j}$ $a_{G}^{i}$

$x_{l}^{j}$ : the current context state of the user contexts $\forall l = 1, 2, \dots, L$

From CRM KB:

$π_{l}$ : the set of possible sensor groups $G$ for each of the requested L contexts

$S^{m}$ : the available sensors

$E^{m}$ : the energy consumption by each sensor $S^{m}$

From LUT:

$E_{G}^{j}$ : the pre-computed energy consumption for each group of sensors $G$ in recognizing context state $x_{l}^{j}$

$D_{G}^{j}$ : the pre-computed delay of each sensor group $G$ in recognizing context state $x_{l}^{j}$

$a_{G}^{i}$ : the pre-computed optimized sensing schedules for the sensor groups in $π_{l}$ in recognizing each context state.

Output: $y_{G}$ , $a_{G}^{i}$

$y_{G}$ : The boolean variable representing the selected sensor groups to recognize multiple contexts states $x_{l}^{j}$

$a_{S^{m}}^{i}$ : Sensing schedules corresponding to the sensors of the selected group including the synchronized schedules

1:
Derive all the possible union of sensor groups $\cup_{l} G$ available in $π_{l}$
2:
for each $\cup_{l} G$ do
3:
for each $S^{m} \in \cup_{l} G$ do
4:
if $S^{m} \in G$ then
5:
Assign $a_{G}^{i}$ obtained from the LUT to $S^{m}$ of group $G$ as $a_{S^{m}}^{i}$
6:
else if $S^{m} \in G_{w}^{l}$ & $G_{q}^{l}$ for $w \neq q$ then
7:
Synchronize sensing schedules $a_{G_{w}^{l}}^{i}$ & $a_{G_{q}^{l}}^{i}$ and assign to $S^{m}$ as $a_{S^{m}}^{i}$
8:
end if
9:
end for
10:
Calculate objective function values according to equation (2)
11:
Track sensors $S^{m} \in \cup_{l} G$ and sensing schedules $a_{S^{m}}^{i}$ with minimum objective function value according to equation (2)
12:
end for
13:
Return sensors $S^{m} \in \cup_{l} G$ with $y_{G} = 1$ and the corresponding sensing schedules $a_{S^{m}}^{i}$ having the minimum objective value
14:
if state $x_{l}^{j}$ changes to $\bar{x_{l}^{j}}$ then
15:
Record time spent in changed context state $x_{l}^{j}$
16:
end if