Wireless Sensor Network Coverage Optimization: Comparison of Local Search-Based Heuristics

Table 2

Selected papers on MLCP optimization with Non-Disjoint Set Cover based approaches—Part I.

No.	Reference	Year	Brief information about the approach and its optimization goal
1	P. Berman et al. [23]	2004	For a given series of covers, the method maximizes the sum of activation times of the covers by formulating the problem as a packing linear program
2	M. Cardei et al. [13]	2005	(1) The problem of set cover maximization is modeled and solved by the mixed integer programming (MIP), and then a linear programming based heuristic computes the covers based on the solution returned by MIP. (2) A greedy heuristic solves the problem of set cover maximization
3	D. Zorbas et al. [24]	2010	A greedy heuristic reinforced by the idea of critical targets (POIs with small neighbor-sensor sets) finds the maximum number of covers which is equal to the schedule makespan since sets provide coverage for the same unit of time
4	K. Deschinkel [25]	2011	The column generation based heuristic finds covers and maximizes the sum of activation times of the covers for the problem modeled with a linear programming formulation
5	Manju and A. K. Pujari [26]	2011	High-energy-first heuristic finds covers using the highest remaining battery states; the selection priorities change after every execution of the selected set, hence the variety of sets over the network lifetime
6	H. Mohamadi et al. [27]	2013	Three learning automata-based scheduling algorithms find both disjoint and non-disjoint covers and then optimize their cardinality
7	A. Tretyakova and F. Seredynski [28]	2013	Two approaches—genetic algorithm and memetic algorithm with a two-dimensional representation of schedules (columns: covers; rows: sensors)—maximize the number of feasible covers satisfying the battery capacity restriction in the rows
8	F. Castaño et al. [29]	2014	The problem of set cover maximization is solved by the column generation based framework having embedded the greedy randomized adaptive search procedure and the variable neighborhood search
9	A. Tretyakova and F. Seredynski [7]	2015	Simulated annealing with a two-dimensional representation of schedules (columns: covers; rows: sensors) maximizes the number of feasible covers satisfying the battery capacity restriction in the rows
10	Y. E. E. Ahmed et al. [30]	2016	A genetic algorithm maximizes the number of covers and schedules them for makespan maximization
11	Y. E. E. Ahmed et al. [31]	2016	A genetic algorithm with problem-specific operators maximizes the number of covers and schedules them for makespan maximization
12	A. Tretyakova et al. [8]	2016	Graph cellular automata approach for cover generation and makespan maximization where cells correspond to sensors, and the neighborhood relation between cells maps the existence of targets or areas commonly monitored by these sensors
13	K. Trojanowski et al. [4]	2017	Local search-based approach with problem-specific perturbation operators (LS) for cover generation and makespan maximization