Receiver alignment dependence of a GA controlled optical wireless transmitter

Abstract

Abstract. A genetic algorithm controlled multispot transmitter is demonstrated to be capable of optimising the received power distribution for single element receivers in fully diffuse mobile indoor optical wireless systems. By dynamically modifying the intensity of individual diffusion spots, the transmitter is capable of compensating for changes in receiver alignment, user movement and surface reflectivity characteristics, with negligible impact to bandwidth and RMS delay spread. The dynamic range, referenced against the peak received power, can be reduced by up to 27% when the room is empty, and up to 26% with user movement and variable receiver alignment. Furthermore, received power perturbation, induced by user movement, is reduced from 10% to 2.5%. This method shows potential for providing a highly adaptable solution of overcoming channel variability whilst also reducing receiver complexity.

Multi-user indoor optical wireless communication system channel control using a genetic algorithm

“…This initial work contained a major simplification in an assumption that the receivers were all vertically orientated, even whilst moving. A further study extended the ability of the GA to allow for user-induced random alignment variability with good success [16]. However, as previously mentioned, this GA technique becomes beneficial when the number of receivers becomes larger, for which, in this paper, the feasibility is shown to extend to multiple users, each with different movement patterns in multiple environments.…”

“…Each receiver has a position vector r R j , orientation vectorn R j , active optical collection area A R j and a field of view FOV R j , defined as the maximum uniaxial symmetric incident angle of radiation with respect ton R j that will generate a current in the photodiode. Furthermore, according to previously published work into the effects of mobile receiver alignment statistics on system performance [16], the orientation in the x and y axis of each receiver is derived from a normal distribution with meanz = 0 (no rotation), and standard deviation σ = 11.7, providing a respective 0.8 probability of rotation within ±15 • , and 0.99 within ±30 • from the unrotated case in each axis.…”

“…In that paper, 2 algorithms, derived from over 200 possible permutations of the algorithms sub-routines was found to be suitable for this type of optimisation scenario. Furthermore, the 2 proposed algorithms were also shown to be successful when applied to incrementally more complex scenarios including random user alignment of the receivers [16]. The work presented here uses the same 2 GAs carefully developed there, so in the interest of conciseness, only a brief factual description will be provided here.…”