Optical wireless communication has been extensively investigated for its potential to expand the transmission capacity and resolve the wireless traffic problem. Challenges occur in terms of free-space optical communication when opaque obstacles occlude the field of view or atmosphere scatters the signal, which makes it hard to align the transmitter and receiver. Non-line-of-sight (NLOS) communication has consequently emerged as a candidate for both indoor and outdoor scenarios, relieving the acquisition requirement while maintaining the transmission rate. To further expand the channel capacity of NLOS communication, optical orbital angular momentum (OAM), described by theoretical unlimited topological charge , can be considered to add the dimensionality of data transmission. However, a crucial limitation to further extending the channel amount of this scheme is the resolution of OAM modes at the receiver. Even weak turbulence causes considerable problems for communication systems based on OAM, which is more severe in NLOS scenarios due to strong scattering or diffuse reflection process. Such aberrations and diffuse losses critically augment the difficulties of discriminating OAM modes.

Fig. 1 Scenario of OAM-based NLOS communication

Here, we propose and demonstrate an approach to collocate NLOS communication with infinitely dimensional OAM modes. To take full advantage of its bandwidth resources, an intelligent neuron network (NN) algorithm is designed at the receiver to decode information from NLOS light. A multi-receiver broadcast experiment is performed to demonstrate the robustness of the method for being applied in atmosphere and underwater communication, as shown in Fig. 2. Due to the robustness of the intelligent NN algorithm, the bit accuracy of the rainbow image at all three receivers is comparable to the corresponding test set, and the highest one can reach 98.91%. The fluctuation at P2 is caused by its lower valid photon counts because P2 is located at a relatively large reflection angle from the diffuser where the optical power is only -74.42 dBm. The rough phase distortion of the glass diffuser is against the impact of projection measurement, which can be improved is atmosphere and underwater scattering.

Fig. 2 The multi-receivers broadcast experiment to transfer a colorized rainbow image.

Then, we investigate the relation between channel dimensionality and the measurement approach. The histograms in Fig. 3 display the impact on NLOS communication bit accuracy versus the number of OAM channels and the number of projection vectors. Obviously, when the amount of multiplexed OAM channels in decreases, the NN can more precisely measure the topological charge value for transmitted light as a result of the simpler phase distribution. In addition, since the increased number of projected measurement can acquire more information from the light intensity at the receiver, increasing the number of projection vectors can efficiently decrease the bit error rate (BER).

Fig. 3 Relation between the bit accuracy of OAM-based NLOS communication and the number of projection vectors and multiplexed OAM channels.