Application Of Compressed Sensing In Space Division Multiplexing Optical Fiber Communication

Space division multiplexing(SDM)transmission technology is considered for overcoming current limitation of the transmission capacity in a single mode fiber,and mode-division multiplexing(MDM)over a few-mode fiber(FMF)is one of the promising approaches.A major technological challenge to mode multiplexed FMF networks is the mode conversion at either end.There are several potential solutions to this problem using mode selective couplers,photonic lantern,long-period gratings,tapered couplers,and free-space optics.Compare to other approaches,free-space optics can efficiently(de)multiplex signals over a wider wavelength range.Free-space optics are used for mode conversion when linearly polarized(LP)modes are generated by spatial light modulation(SLM).In a recent paper,the authors propose a scheme that achieves selective excitation of an arbitrary LP mode of the parabolic-profile graded-index(GI)FMF by SLM with the binary-phase mask.But this binary-phase mask generates a far-field distribution in the Fourier plane which is inaccurate.It is necessary to generate more accurate phase masks for desired mode field.Genetic algorithm(GA)is a kind of adaptive heuristic search algorithm which simulates the evolutionary process of natural selection,species evolvement,and population genetics.It has been used in a wide spectrum of optimization problem.In this paper,the phase masks are generated using GA,which can create profiles that in theory are identical to the desired modes.It differs from previous binary phase masks that these masks are multi-phase.In order to produce high-quality masks,the algorithm is implemented on a graphics processing unit(GPU)to speed up the computation.We used three types of selection strategy in the GA procedure,and tournament selection is the optimal selection among three operators.Due to the computational complexity brought by high-dimensional signal,compressed sensing is used as a deeper optimization and improvement.Unlike traditional compression methods,compression sensing enables large reduction in the sampling which is below the Nyquist sampling frequency.By analyzing the characteristics of the signal,we choose the most suitable sensing matrix,and achieve the improvement of the time efficiency of the mode conversion.