NOMA for multiple access. GFDM for multi-carrier modulation.

Dr. Haci, I need some reference papers supporting your suggestion please. And also some suggestions for the reasons.

As far as I know, this has not been decided yet in 3GPP. But almost all the companies involved in 3GPP have proposed different forms of OFDM. So it is quite likely that 5G will use OFDM with additional filtering or tweak to better support certain types of traffic.

Please note that the modulation choice (such as OFDM) has little to do with increasing the spectral/energy efficiency in 5G. Higher data rates will be achieved by a combination of MIMO transmission and using more bandwidth.

Dear Emil,

Thanks for the answer. I agree that spectral efficiency will be achived through MIMO and more bandwidth. I also suggest that small-cells and frequency re-use is also a means of improving spectral efficiency.

Yet, I think people should not give up exploring better modulation and multiple access schemes than OFDM. It looks like companies are happy about OFDM and not looking for a competant scheme. I am planning to start a research in this field as well.

The spectral efficiency (measured in bit/s/Hz/cell) is only improved by MIMO. Small cells and more bandwidth will not have any direct effect on the spectral efficiency, but small cells may improve the area spectral efficiency (bits/s/Hz/km2) and more bandwidth might improve the data rates which are computed as bandwidth x spectral efficiency.

OFDM is already quite good. What you can gain by changing it is essentially to remove the cyclic prefix (around 10% improvement in data rate) and to achieve something that is easier to implement. But you need to find something that is really good if it should be worth changing what we already have and now how to build.

I think one thing about Frequency efficiency (and also power efficiency) is the use of underlaid systems such as device-to-device (D2D) communication. Although D2D seems way far from final implementation today. But as it is suggested that D2D communication will be re-using the uplink resources, hence frequency re-use consequently increasing the frequency efficiency. Also, this will consume less power for neighboring d2d devices to communicate together than through the network. It will also reduce communication latency. 

About NOMA...

I don't think NOMA is possible in 5G systems. Since we do not have hardware technology that can practically implement NOMA. Even if we get one on the core-network-side, distributing LARGE number of mobile sets with NOMA enabled technology is far from practicality now.

However, NOMA may appear may be in post-5G systems since it is potentially a great application.

I read the article suggested by Dr. Hacı. It is interesting and helpfull but I could not find it in 3GPP or any other Horizon 2020 framework.

D2D communication can indeed improve the spectral and energy efficiency. If you need a reference, we studied this in the following paper:

Serveh Shalmashi, Emil Björnson, Marios Kountouris, Ki Won Sung, Mérouane Debbah, “Energy Efficiency and Sum Rate Tradeoffs for Massive MIMO Systems with Underlaid Device-to-Device Communications,” EURASIP Journal on Wireless Communications and Networking, 2016:175, 2016.

There is however a fundamental problem with D2D: Will there be any traffic that is suitable for D2D communication? One can think about some niche application in public safety and real-time media distribution in an arena, but will any of these applications be used in practice? If D2D should bring any actual improvements to the spectral and energy efficiency of the 5G systems, we need to have a sufficiently large fraction of the traffic that can be sent as D2D.

NOMA enables multiple users to communicate simultaneously with a single-antenna base station, using the same time and frequency resources. But once we equip the base stations with a large number of antennas, which is expected to be the case in 5G (called massive MIMO), the users can be separated using spatial beamforming rather than complicated NOMA interference mitigation methods. Maybe it can play a role in pre-5G systems, but I don't think it will be of much use in 5G.

OFDMA is widely adopted as the multiple access technique in current wireless communication systems. Although OFDMA approach can eleminate interference among concurrent transmissions in orthogonal subcarriers, it does not generally achieve the highest possible data-rate for a given error probability [1]. Thus, the requirement for higher spectral efficiency and expected improvement of receiver processing capabilities made NOMA [2] a candidate technology for future wireless communications.

Experimentations (by NTT DOCOMO, Japan) for downlink transmissions with real-hardware testbed showed up to 61%, and real-time simulator for very dense urban area (actual model of Shinjuku, Tokyo) showed up to 51%, of throughput gain of NOMA over OFDMA [3]. The promising concept and experimental results made NOMA a study item in Release 13 of the 3rd generation partnership project standardization [4].

[1] T. M. Cover and J. A. Thomas, Elements of Information Theory. New York, NY, USA: Wiley-Interscience, 1991.

[2] R. Zhang and L. Hanzo, “A Unified Treatment of Superposition Coding Aided Communications: Theory and Practice,” IEEE Communications Surveys and Tutorials, vol. 13, pp. 503–520, Third 2011.

[3] NTT DOCOMO, “5G Concept and Technologies.” http:5gworkshop.hhi.fraunhofer.de/wp-content/uploads/2014/12/Globecom-2014-WS-on-5G-New-Air-Interface-NTT-DOCOMO.pdf, December 2014. Accessed:2015-03-17.

[4] 3rd Generation Partnership Project, “Release 13: RP-141831.” http:/www.3gpp.org/release-13, September 2014. Accessed: 2015-03-17.

Huseyin Haci: As you say, NOMA can improve the spectral efficiency over OFDMA with 1.5x or maybe even 2x. But we need to put that into perspective. Massive MIMO technology can improve the spectral efficiency by 20x. This has, for example, been demonstrated earlier this year in Bristol, where 22 users were served by 256-QAM simultaneously:

http://spectrum.ieee.org/tech-talk/telecom/wireless/5g-researchers-achieve-new-spectrum-efficiency-record

Both Hüseyin and Emil are on something valuable. Thanks for the highly helpfull contributions. Huseyin is very favorable for NOMA but there are only a number of articles defending it. No standardisation activity is available yet. We do appreciate the enhancement of performance by MIMO. But what MIMO does is just diversity, nothing to do with the modulation scheme. I do belive that there are alternative modulation and multiple access techniques out there and I am trying to find out about them. The main objectives of 5G are 

improving capacity by a factor of 1000

improving spectral efficiency by a factor of 10

improving power efficiency by a factor of 10 and finally

reducing latency down to less than 1ms.

Cell Dencification, MIMO and Cooperative Channel Estimation and Interference Calculation are expected to do the job. However, in this discussion, I want to concentrate specifically on the modulation and multiple access techniques that can perform better than the current standard OFDM or its derivatives. Please concentrate on this issue.

You might want to investigate filter bank multi carrier (FBMC), which is actually an older form of OFDM, but has some advantages in spectral efficiency (no guard intervals).

As Emil said, no definite decisions have been made yet. The higher bit rates will be more a matter of using advanced MIMO and of using much higher frequency bands (hence wider channels) than anything magical about "modulation." The reduced latency will be a matter of distributing servers to the network edge, and of avoiding the use of interleaving, rather than anything about changing the laws of physics. Guess what? Even 5G cannot change that c = 3E8 m/sec.

When one sees the goals of any such new systems, they always sound like magic. My preference is to explain what is behind these goals, rather than attempting to make people's jaws drop in amazement.

Thanks for the reminder Albert. I appreciate it. I agree that OFDM was a magical innovationnin terms of spectral efficiency and power saving. The cooperative channel estimation and successivevinterference calculation vould also make significant contribution but where is the limit? Where is the actual benchmark to the spectral efficiency? Is there still room forvimprovement? This discussion realky excites me! Other researcheers who are dealing with modulation and multiple access schemes. Please join in and let us have your contribution. 

Hello Emil, my taught for future wireless networks is as follows; It is hard for a single technology to improve the system performance, e.g. the data-rate, to meet the performance requirements in future wireless communications. Performance gains of multiple technologies, working in harmony, need to be combined to realize the extra high data-rate communications. NOMA is an advantageous technology in this aspect as well. Since it can work in conjuction with various air interface technologies, such as multiple-input multiple-output (MIMO) and distributed antenna systems, and networking platforms, such as small cells, proposed for future wireless communications without the need for any extra hardware. Employing NOMA on top of another technology will provide the additional frequency reuse gain to further improve the performance. An interesting application scenario of NOMA is envisioned to be used in conjuction with beamforming technology in the small cell environment [1]. Please refer to [2] for further details on the application of MIMO to NOMA.

[1] Y. Saito, Y. Kishiyama, A. Benjebbour, T. Nakamura, A. Li, and K. Higuchi, “Non-Orthogonal Multiple Access (NOMA) for Cellular Future Radio Access,” in 2013 IEEE 77th Vehicular Technology Conference (VTC Spring), pp. 1–5, June 2013.

[2] Z. Ding, F. Adachi, and H. Poor, “The Application of MIMO to Non-Orthogonal Multiple Access,” IEEE Transactions on Wireless Communications, vol. PP, no. 99, pp. 1–1, 2015.

I dont think there is anything about NOMA in the horizon for 5G as Huseyin suggested. The case of MIMO is also nothing new. We have been using space diversity at both the transmitter and receiver to produce multiple statistically independent copies of the original message at the receiver. What Albert suggests regarding MIMO is important bu in simple words, what is MIMO other than exaggerated use of space diversity and using large number of antenna elements or sensors as some researchers call them. 

I like what Albert says about the laws of physics but the same laws of physics were there when we switched from analog to digital, from single user to multiuser CDMA, from CDMA to OFDM etc. I still belive that there are modulation and multiple access schemes to discover but I dont think it is NOMA.

Huseyin Haci: You can certainly utilize successive interference cancelation schemes, such as NOMA, on top of other multiple access technologies. But in a Massive MIMO system, the many antennas make the user channels nearly orthogonal so there is not much to gain from successive interference cancellation. (See Myth 5 in my paper "Massive MIMO: Ten Myths and One Critical Question").

Hasan Amca: The Shannon capacity is one fundamental limitation. The capacity of a SISO communication channel is well understood and there are practical ways to operate close to the capacity. Since we cannot beat the capacity, what we need to do is to create more parallel links that together increase the sum capacity. This is what MIMO is all about. Massive MIMO is indeed a variation on traditional multi-user MIMO theme, but it is very nice because it both makes the gains larger and easier to achieve in practice.

 am reading your paper "Massive MIMO: Ten Myths and One Critical Question" Emil. Thanks...

Since we talked about spectral efficiency, you might be interested in my recent blog post "How much does Massive MIMO improve the spectral efficiency?"

http://www.massive-mimo.net

I read both articles but I have not seen. Anything other than the good old antenna diversity. Now I will post a question on this.