Low-Complexity Bit Allocation for RSS Target Localization

The overall localization error performance for a sensor network employing quantized received signal strength (RSS) measurements largely depends on the number of bits used and their distribution. In this paper, we consider a rate-constrained bit allocation problem in which multiple points of interest (PoIs) are assigned possibly non-uniform error requirements. The main objective is to allocate a number of bits in order to satisfy error requirements as best as possible in a squared error sense. Due to the computational complexity of optimally solving the allocation problem, two novel low complexity allocation algorithms are proposed. The main idea behind these algorithms is to relate localization errors to corresponding bit density requirements. In the first algorithm, bit allocation is formulated as a sequential constrained convex optimization problem which can be solved with efficiency using available convex solvers. The second algorithm is a greedy version of the first algorithm in which sensors are selected for allocation based on the bit density differences at their neighboring PoIs. Numerical experiments show the efficacy of the proposed algorithms in comparison to other available algorithms in terms of overall error, coverage, and computational complexity.