Michał Maternia (Nokia)
In recent years, wireless communication systems exploit higher and higher frequency ranges, in order to find necessary radio resources to serve the raising traffic demand. Some WRC-15 bands selected for studies in the context of 5G, as well as the 28 GHz band chosen in large markets for 5G deployments, span over hundreds of MHz and allow straightforward handling of bandwidth hungry broadband services . However, propagation of radio waves in millimetre waves region, differs strongly from radio propagation at traditional cellular frequencies. Firstly, radio signals face much higher path losses and, secondly, the role of propagation phenomena such as, diffraction or reflection for most materials, is much less prominent. This leads inevitably to exploitation of massive MIMO antenna systems at millimetre waves bands.
As described in previous blog entries (cf. MIMO techniques and architectures for millimetre wave mobile communications), massive MIMO antennas that exploit beamforming are used at higher frequencies to cope with increased path losses. A critical factor for wideband equipment is the power consumption of digital-to-analog converters (DAC) that scales with the sampling rate and the number of bits per sample. For this reason, at higher frequencies analog beamforming solutions with low number of DACs are preferred over digital beamformers that require a separate DAC for each processed transmission stream. After analog beamforming, the radio beam offers high antenna gain (tens of dB) and is much narrower (several degrees for 3 dB beam width) comparing to the output of contemporary sector antennas. This narrowness brings several novel system level design implications that were not present in the previous cellular generations.
Figure 1: Operations with sector antennas (left) and mMIMO with analog beamforming (right)
During transition from idle to connect mode (e.g., when we want to use some service after longer inactivity period) the radio network (gNB) and user equipment (UE) need to determine a spatial direction of the suitable communication link, which boils down to the selection of one out of several potential radio beams (so called P-1 procedure ). In order to facilitate this selection, 3GPP specify details of selected radio time slots that will be used by gNB to sweep several spatial directions with transmissions of so called synchronization signal blocks. In each block (spanning over 4-6 OFDM symbols), a transmission over one beam direction will consist of synchronization signals and broadcast information that needs to be obtained before exchange of initial access information . In other slots that are used to detect initial access messages (Random Access Chanel (RACH)), gNB will tune it’s receive antennas to sweep receive beams. If there is a fixed time relation between the transmission and reception at a given beam, the UE will be able to calculate it based on system information broadcasted in synchronization signal blocks, and will use appropriate timing for transmission of initial access preamble to indicate suitable radio beam. Alternative solutions are also possible, e.g., in the case of no beam correspondence at the gNB, UE may repeat the initial access preamble for several transmission opportunities.
In previous cellular generations, UE had to monitor radio signals from neighbouring radio cells to facilitate potential switch of the serving cell. When the radio signals are limited to narrow beams a different approach is needed. For start, 3GPP is working on the new procedures for beam tracking/refinement at the gNB side caused by e.g., UE movement (P-2) or tracking/refinement of UE beams that is needed e.g. due to UE rotation (P-3) . Further on, instead of reporting measurements for the best cells, UE will report measurements for a number of best beams that were detected. Additional standardization efforts are put for development of a mechanisms that will be used to recover after beam failure:
- detection of the beam failure,
- identification of the new beam candidate,
- beam failure recovery request transmission, and
- response for the beam failure recovery request.
Due to the high cost of radio processing chains of gNBs operating at higher frequencies, digital MIMO operations (including higher order MIMO i.e. transmission of multiple data streams) and frequency division multiplexing of the scheduled users, are not straightforward. Therefore the most accessible scheme is a beamformed transmission toward single user in a single slot. To enable efficient resource and interference management in 5G, 3GPP work to design of specific reference signals and feedback types. Because of a high directivity of radio transmissions, a cross link interference mitigation is needed and in 5G at least the information of the intended UL/DL transmission direction configuration is exchanged at the backhaul (other methods, including gNB-gNB and UE-UE measurements, are being investigated ). When operating at higher frequencies, UE need to track specific reference signals for phase tracking, to avoid additional phase noise errors resulting from drifts in the local oscillators. Finally, as cloud deployments are gaining more and more traction, more centralized scheduling mechanisms will also come into play in 5G. This is also reflected in 3GPP decision for the split of gNB into centralized and distributed unit .
 METIS-II D3.1 “5G spectrum scenarios, requirements and technical aspects for bands above 6 GHz”, ICT-671680 METIS-II Deliverable 3.1, Version 1.
 3GPP, Study on New Radio Access Technology Physical Layer Aspects (Release 14), 3GPP TR 38.802, March 2017.
 3GPP, Summary of discussion on SS block composition, SS burst set composition and SS time index indication, 3GPP TDoc R1-1706534, April 2017.
 3GPP, WF on Framework of Beam management’ 3GPP TDoc R1-1703523, February 2017.
 3GPP, WF on cross link interference mitigation enablers, 3GPP TDoc R1-1706222, April 2017.
 3GPP, Study on new radio access technology: Radio access architecture and interfaces (Release 14), 3GPP TR 38.801, March 2017.