Positioning via direct localisation in C-RAN systems

What is it about?

Cloud radio access network (C-RAN) is a prominent architecture for fifth generation wireless cellular system that is based on the centralisation of baseband processing for multiple distributed radio units (RUs) at a control unit (CU). In this study, it is proposed to leverage the C-RAN architecture to enable the implementation of direct localisation of the position of mobile devices from the received signals at distributed RUs. With ideal connections between the CU and the RUs, direct localisation is known to outperform traditional indirect localisation, whereby the location of a source is estimated from intermediary parameters estimated at the RUs. However, in a C-RAN system with capacity limited fronthaul links, the advantage of direct localisation may be offset by the distortion caused by the quantisation of the received signal at the RUs. In this study, the performance of direct localisation is studied by accounting for the effect of fronthaul quantisation with or without dithering. An approximate maximum likelihood localisation is developed. Then, the Cramér–Rao bound on the squared position error of direct localisation with quantised observations is derived. Finally, the performance of indirect localisation and direct localisation with or without dithering is compared via numerical results.

Featured Image

Why is it important?

The C-RAN architecture, with its centralised baseband processing at a CU, enables the implementation of direct, or one-step, localisation. While direct localisation outperforms the traditional indirect, or two-step, localisation in the presence of ideal connections between the CU and the RUs, this may not be the case in the C-RAN due to the distortion caused by quantisation on the fronthaul links. In this paper, we have studied the performance of direct and indirect localisation in the C-RAN by means of the calculation of the CRB as well as quantisation effect on the localisation accuracy via analytical and numerical results. Interesting open problems concern the study of the optimal scalar quantisation design for the RUs to maximise the localisation accuracy.