Developing mmWave mobile radio interface
mmWave mobile communication is challenging due to harsh radio propagation conditions and severe hardware impairments that are experienced at extremely high carrier frequencies and large signal bandwidths. Ericsson Research, in partnership with leading research organizations in Europe, has led the development of radio interface for mmWave mobile communications. Let’s have a look at this innovative design.
A large amount of spectrum is available in the mmWave frequency band (30 – 300 GHz). However, there is no commercial mobile communication system operating in mmWave frequencies today – LTE is designed for frequencies below 6 GHz. There are only local area networks and (mostly) indoor communication systems based on the IEEE 802.11ad and 802.15.3c standards operating in the unlicensed 60 GHz band. IEEE 802.11ay, a follow-up of 802.11ad, is under development. 3GPP is currently working towards a global standard for 5G New Radio (NR) that will operate in frequencies from below 1 GHz up to 100 GHz. NR aims to unleash mmWave frequencies to boost network capacity and deliver ultra-high throughput services to large number of mobile users before 2020.
Here’s an illustration of operational frequency ranges of existing (2G, 3G, 4G) and future (5G) mobile communication systems:
Frequency ranges of current and future mobile communication systems
As a way of accelerating the NR standardization process, mmMAGIC project (co-funded by the European Commission’s H2020 program) has developed and validated a novel mmWave radio interface. The radio interface is flexible to support carrier frequencies up to 100 GHz under both standalone and non-standalone operations. Parts of this radio interface have already been adopted for NR (agreed in 3GPP).
The key technology components of the mmWave radio interface are summarized below.
OFDM waveform has been selected (among several waveforms) due to its high spectral efficiency, flexibility, MIMO compatibility, and low implementation complexity. OFDM waveform with flexible numerology (subcarrier spacing, cyclic-prefix) addresses wide range of frequencies and deployments, service latency and reliability requirements, user mobility, and hardware impairments. Optional OFDM features include low complexity receiver agnostic processing to improve spectral confinement and peak-to-average power ratio, advance prefix design for channel estimation, and DFT precoding for enhancing power efficiency.
On adoption of OFDM for 5G NR, see our blog post In the race to 5G, CP-OFDM triumphs!
2. Channel codes
LDPC, Polar, and Turbo codes have been compared for throughput, throughput per chip area, complexity, error correction performance, latency, and suitability to retransmissions for both short and long code block lengths. Based on comparisons, LDPC codes are selected for high throughput and low latency data transmission. Polar codes are adopted for transmission of reliable control information. Moderate complexity channel decoders have been developed for ultra-high throughput transmission.
3. Re-transmission schemes
Adaptive and asynchronous HARQ has been adopted to provide fast and reliable retransmissions in single-hop and multi-hop scenarios (e.g., wireless backhauling). HARQ has a design flexible to tradeoff between reliability and latency, depending on the scenario. Optional features include techniques for further reducing latency, for example, retransmissions without full FEC decoding.
4. Frame structure
A common frame structure for TDD and FDD that supports fast decoding of data via front loaded reference symbols, fast acknowledgements for retransmissions, integrated access and backhaul transmissions, transmission in both licensed and unlicensed spectrum, and slots of varying lengths for different types of services (e.g., shorter slot durations for latency critical applications and longer slot durations for services with less delay stringent requirements to reduce reference signal overhead).
5. Reference signals
Beamformed reference signals for phase noise compensation, channel estimation for data demodulation, and channel state acquisition for MIMO transmission in both UL and DL. Reference signals are designed to optimize network energy efficiency.
6. Duplexing schemes
Primarily TDD due to following reasons: i) the mmWave spectrum is mainly unpaired; ii) guard period (GP) for link direction (UL/DL) switching can be kept small for small cells which are in focus for mmWave deployments; iii) TDD operation makes the UL and DL channels reciprocal, which is important for massive MIMO operation (to estimate DL channel via uplink channel state information); iv) DL/UL traffic can be very dynamic. Dynamic TDD matches the DL/UL traffic load of each small cell individually by adjusting the GP position.
7. Multiple acces schemes
Primarily, SDMA due to extensive use of beamforming in mmWave. The following restrictions exist due to hybrid beamforming: The number of links to be multiplexed via spatial domain is limited by the number of RF chains of the transceiver, and no subcarrier dependent precoding is possible (e.g. due to constraint of analog phase shifters). Accordingly, TDMA should be used on top to serve different SDMA groups (each group containing several links under SDMA). FDMA can be used further on top to multiplex users with the same/similar beam directions.
8. Initial access schemes
The initial access procedure starts with a cell discovery phase, followed by random access phase, during which network resources are allocated to the user, and conclude with a beam refinement/tracking phase. For the cell discovery, different beam sweeping options (time division, frequency division, code division and space division) exist. A novel random access preamble (based on repetitions of short OFDM symbols) improves energy efficiency, reduces latency, and implementation complexity. For the beam refinement/tracking phase, novel schemes based on genetic algorithm and pseudo-exhaustive/probabilistic methods are employed to reduce the latency in scenarios with mobile users.
9. Dynamic spectrum sharing
Network operators can reuse/share spectrum by exploiting separation in spatial domain (to reduce interference) via beamforming. Various spectrum sharing architectures (based on different level of coordination between RAN and core network) and supporting functions (spectrum sensing, enhanced CSI acquisition and exchange, synchronization, beam coordination) have been developed and evaluated, showing significant benefits of dynamic spectrum usage.
Watch a field trial of mmMAGIC radio interface here.
The mmMAGIC consortium comprises of 18 organizations in Europe (Samsung, Ericsson, Huawei, Nokia, Alcatel-Lucent, Intel, Orange, Telefonica, Keysight, Rhode & Schwarz, HHI, CEA-Leti, Imdea Networks, Bristol University, Chalmers University, TU Dresden, Qamcom, Aatlo University). For details on the mmWave radio interface design, please visit project webpage: https://5g-mmmagic.eu/