Digital Processing for an Analogue Subcarrier Multiplexed Mobile Fronthaul

In order to meet the demands of the fifth generation of mobile communication networks (5G), such as very high bit-rates, very low latency and massive machine connectivity, there is a need for a flexible, dynamic, scalable and versatile mobile fronthaul. Current industry fronthaul standards employing sampled radio waveforms for digital transport suffer from spectral inefficiency, making this type of transport impractical for the wide channel bandwidths and multi-antenna systems required by 5G. On the other hand, analogue transport does not suffer from these limitations. It is, however, prone to noise, non-linearity and poor dynamic range. When combined with analogue domain signal aggregation/multiplexing, it also lacks flexibility and scalability, especially at millimetre wave frequencies.

Measurements (matched in simulation) of analogue transport at millimetre wave frequencies demonstrate some of these issues. High data rates are demonstrated employing wide bandwidth channels combined using traditional subcarrier multiplexing techniques. However, only a limited number of channels can be multiplexed in this manner, with poor spectral efficiency, as analogue filter limitations do not allow narrow gaps between channels.

To this end, over the last few years, there has been significant investigation of analogue transport schemes combined with digital channel aggregation/ de-aggregation (combining/ separating multiple radio waveforms in the digital domain). This work explores such a technique.

Digital processing is used at the transmitter to flexibly multiplex a large number of channels in a subcarrier multiplex, without the use of combiners, mixers/ up-converters or Hilbert transforms. Orthogonal Frequency Division Multiplexing (OFDM) - derived Discrete Multi-Tone (DMT) and Single Sideband (SSB) modulated channels are integrated within a single Inverse Fast Fourier Transform (IFFT) operation. Channels or channel groups are mapped systematically into Nyquist zones by using, for example, a single IFFT (for a single 5G mobile numerology) or multiple IFFTs (for multiple 5G mobile numerologies).

The analogue transport signal generated in this manner is digitally filtered and band-pass sampled at the receiver such that each corresponding channel (e.g. channels destined to the same radio frequency (RF)/ millimetre wave (mmW) frequency) in the multiplex is presented at the same intermediate frequency, due to the mapping employed at the transmitter. Analogue or digital domain mixers/ down-converters are not required with this technique. Furthermore, each corresponding channel can be readily up-converted to their respective RF/mmW channels with minimal per-signal processing. Measurement results, matched in simulation, for large signal multiplexes with both generic and 5G mobile numerologies show error-vector magnitude performance well within specifications, validating the proposed system.

For even larger multiplexes and/or multiplexes residing on a higher IF exceeding the analogue bandwidth and sampling rate specifications of the ADCs at the receiver, the use of a bandwidth-extension device is proposed to extend the mapping to a mapping hierarchy and relax the analogue bandwidth and sampling rate requirements of the ADCs. This allows the receiver to still use digital processing, with only minimal analogue processing, to band-pass sample smaller blocks of channels from the larger multiplex, down to the same intermediate frequency. This ensures that each block of channels is within the analogue bandwidth specification of the ADCs. Performance predictions via simulation (based on a system model matched to the measurements) show promising results for very large multiplexes and large channel bandwidths.

The multiplexing technique presented in this work thus allows reductions in per-channel processing for heterogeneous networking (or multi-radio access technologies) and multi-antenna configurations. It also creates a re-configurable and adaptable system based on available processing resources, irrespective of changes to the number of channels and channel groups, channel bandwidths and modulation formats.