At last month’s Mobile World Congress it was good to see a few demonstrations of 5G air interface technologies. As is typical in this phase of R&D, these demos were limited to big boxes with a science project feel. At one such demo at the Intel booth a characteristic oscilloscope rendering of an OFDM waveform begged the question what role will OFDM play in 5G? What waveform will define 5G? Even, will 5G say a farewell to OFDM?
While a definitive answer is not agreed yet, there is no doubt that the journey of insights to one conclusion or another has already started. The first insights are already coming from the three key 5G use cases being tackled by 3GPP, namely, enhanced Mobile Broadband (eMBB), massive Machine Type Communications (mMTC), and ultra-reliable low latency communications (URLLC).
Each of these scenarios puts the priority on certain Key Performance Indicators (KPI), for example, increased network capacity and higher peak data rate for eMBB, connection density and energy efficiency for mMTC, and high reliability and low latency for URLLC. Put together, it is fairly obvious that these use cases and their KPIs result in a wide set of challenging requirements that the 5G air interface design will have to cope with.
+ ALSO ON NETWORK WORLD 5G is coming and it is the future of mobile +
With focus on the waveform, the challenge of supporting the above use cases with their respective KPIs translates into a design challenge on several fronts. These include for example: Peak-to-Average-Power Ratio (PAPR – a measure of energy efficiency), Out-of-Band leakage (OOB - a measure of spectral efficiency), Bit-Error-Rate (BER) in multipath (a measure of reliability), Complexity (at the transmitter and the receiver), Multi-user support, multi-antenna (MIMO) support, Latency, Asynchronicity, etc. It is obviously very challenging to find an optimal design that can achieve the best trade-off between these aspects. For example, it is well known from 4G that multi-carrier OFDM is advantageous compared to single carrier for throughput and complexity, but it becomes disadvantageous in other aspects like PAPR, OOB and asynchronicity.
It is much of the same story repeating now in 5G. Several waveforms are being researched as potential candidates for the new non-backwards compatible 5G air interface, spanning a wide range from very simple single carrier to very sophisticated multi-carrier waveforms. The main candidates include: Single-Carrier Frequency Division Multiplexing or called differently Cyclic Prefix (CP) DFT-Spread-OFDM (as in 4G LTE Uplink), Zero-Tail (ZT) or Unique-Word (UW) DFT-Spread-OFDM, GFDM, UW-OFDM, CP-OFDM (as in 4G LTE Downlink), Resource-Block-Filtered OFDM, Universal Filter Multi-Carrier (UFMC), and Filter-Bank-Multi-carrier (FBMC).
As the list above shows, many of the 5G candidate waveforms still relate somewhat to OFDM. These include CP/ZT/UW DFT-Spread-OFDM (single carrier) and UW/CP/RBF-OFDM (multi-carrier). The other (non-OFDM-related) ones such as GFDM, UFMC and FBMC, are certainly exciting as out-of-the-box new developments. However, at least for GFDM and FBMC, they fall far behind the OFDM-variants in particular on key features such as complexity, latency and maturity of implementation. This tells us clearly that OFDM will certainly remain as the root framework for the new 5G waveform design in both the downlink and the uplink, with some optimization to support the new 5G use cases.
So what kind of optimization are we looking at for the OFDM-based waveform(s) in 5G? First and foremost it is the flexibility and scalability of the OFDM numerology (defined as the OFDM configuration in terms of sub-carrier spacing, symbol duration, cyclic prefix, resource block size, etc.). Unlike 4G, in 5G we are looking into the simultaneous support in the same band of various services with their diverse QoS requirements, hence requiring different numerologies. We are also looking into different spectrum bands (e.g. below and above 6 GHz) and different bandwidths (from MHz to GHz) in addition to different deployment scenarios, which all require scalable numerologies.
Next, especially for mMTC service scenario, the 5G waveform design will need to add means or tools to relax the tight synchronization requirement in the OFDM of 4G. Furthermore, optimizations will be necessary for efficient support of anticipated large MIMO configurations, and in order to lower the implementation complexity especially for wide bandwidths (e.g. above 250 MHz). This is in addition to consistently looking at increasing spectral efficiency and reducing the OOB emissions.
So, to conclude, it is fairly clear that 5G will not say farewell to OFDM but rather evolve the OFDM of 4G into a flexible and scalable waveform framework optimized to cater for all the various usage scenarios envisaged. Couple this with the reality that LTE still has a long life ahead of it. 3GPP recently created the new marketing marker, “LTE-Advanced Pro” for all things Release 13 and beyond. This beyond will very likely form a significant component of what 3GPP submits as its proposal to meet the IMT2020 requirements (aka 5G) in 2017/18.
So long live OFDM! It is going to be with us for a long time to come.