India's 5G Technology - Opportunities and Challenges

Since 2016, India has been trying its best to make contributions to the process of the global 5G technology development process. In alignment with this aspiration, it presented a concept titled low mobility large cell (LMLC) in the ITU meeting held in Geneva in Oct 2016 (WP 5D #25). This proposal was further sharpened (evaluation configuration & spectrum efficiency targets) in a follow-up meeting held in the same venue in Feb 2017 (WP 5D #26). Subsequently, these contributions got included in the final ITU documents - M2410, M2411 & M2412 at the end of 2017. These ITU documents lay the foundation for submission and evaluation of prospective 5G technologies proposed by stakeholders. India, through TSDSI, also submitted its technology, which later got evaluated by various stakeholders for suitability to be considered as 5G technology. The purpose of this note is to identify the opportunities and challenges being faced by India while it goes through this process.

What is LMLC?

In order for us to appreciate India's current situation, we need to first understand what LMLC is all about. It is nothing but a toned-down requirement (configuration "c") defined for evaluation in M2412 for rural eMBB settings only for the 700 MHz band (pages 23 and 24). The idea is to stretch the intersite distance (ISD) much beyond (6 kms) than what is possible under the conventional setting (1.732 kms) in the same band (configuration "a"). This is made possible by diluting some performance requirements like UE speed of interest (from 500 km/h to 30 km/h), thereby enabling stretching of the coverage of BTS compared to its counterpart (configuration "a"). By doing so the cost of deployment can be reduced significantly. Surprisingly, while doing so, the 5th percentile user spectrum efficiency number (evaluating network performance at the cell edge) was done away with. And for the evaluation purpose, only average spectrum efficiency (50th percentile number) was relied upon (table 1 page 4 of M2410). Here, it also important to point out that no discussion happened on 3.5 GHz band for LMLC at the time when evaluation, configuration, and spectrum efficiency numbers for LMLC were being frozen. Hence, one has to conclude that LMLC with 6 Kms ISD (defined in ITU) was only for the 700 MHz band and NOT for any other band.

Integration with 3GPP

After having incorporated configuration for LMLC in the ITU, it was necessary for India to ensure that its proposed technological changes strongly integrates with that of 3GPP. This was necessary for ensuring interoperability, roaming and enabling economies of scale. As the changes suggested by India were only incremental, it had no other option but to use the remaining specification being defined by 3GPP. India did approach 3GPP with its request but it was unable to convince 3GPP to support the proposed changes. Hence, to increase its leverage with 3GPP, it tried to approach ITU with an intent that it facilitates the change of the Global Core Specification (example 3GPP) by the respective transposing organization (TSDSI in this case), which gets done outside the body maintaining the GCS. This was to ensure that the changes in the 3GPP core specifications done by TSDSI sustain in all the future releases of 3GPP 5G standards - a must for supporting interoperability and roaming. But the proposal was not discussed formally in the ITU, and therefore did not go through. The reason - it was practically impossible to implement. How can 3GPP ensure compatibility of fundamental changes done to its own specification by hundreds of SDOs at arm's length? It would have completely destroyed 3GPP's flexibility to maintain and evolve its specification with new features and capabilities in future releases. India also tried requesting 3GPP to reserve bits in its specification to be used by TSDSI in all its future releases and the request was politely turned down by 3GPP. This is due to the challenges in supporting and maintaining changes done outside 3GPP.

India's Own RIT

Given this reason, India had no other option but to propose its own RIT (Radio Interference Technology) for 5G. It did so in the Brazil meeting (WP 5D #32) held in July 2019. But it was asked to resubmit as its proposals did not fully fit the template and requirements as defined in the earlier frozen documents M2410, M2411 & M2412. Hence, India made a revised submission in Sept 2019. This document was nothing but a self-evaluation of the requirement spelled out in the above-mentioned ITU documents. Note that India's submission contained section 20 (page 55) which evaluated an LMLC configuration for 3.5 GHz band which was neither discussed nor agreed upon at the time of finalization of LMLC requirements.

Observation of ITU on TSDSI's RIT

ITU's in its report embedded the following in Dec 2019.

"For submission TSDSI incorporated in Document 5D/1301, it is noted that the provided supplementary material included with the submission as indicated below does not provide information that is directly relevant and pertinent to the IMT-2020 evaluation and does not align with the provision of Report ITU-R M.2411 Section 5.2 to “provide a further understanding of the submission”. WP 5D has observed that this supplementary information, in the parametric values or other assumptions and analysis utilized, does not align with that specified in Report ITU-R M.2412 for a specific scenario being assessed."

"-WP 5D has not considered the indicated supplementary materials in the IMT-2020 evaluation as it is not directly relevant to the formal IMT-2020 evaluation.

– WP 5D, therefore, offers no endorsement of this supplementary information in the context of IMT-2020 suitability."

The supplementary material that is referred to by the ITU report is section 20 of the TSDSI's self-evaluation document which included performance numbers of LMLC of the 3.5 GHz band.

LMLC in 3.5 GHz band

If LMLC was defined only in the 700 MHz then what prompted India to make a submission in the 3.5 GHz band? One can only speculate on the reasons, and I think it may be the following.

a) 700 MHz band is an FDD band and India's PI/2 BPSK (core technology to support LMLC) works best in a TDD band, as TDD enables the UE peak power to be raised to 26 dBm without compromising the SAR values as the average power can still be maintained at 23 dBm.

b) 700 MHz band if operated in TDD mode will make the band useless. The reasons are explained by me in the embedded video.

c) 700 MHz band in India is valued at 8.3 billion USD for 10 MHz blocks of spectrum, thereby making band totally uneconomical for rural usage (gains derived from LMLC will get totally nullified by the cost of the band).

d) 3.5 GHz band is a TDD band where PI/2 BPSK can be used to raise peak power to 26 dBm by keeping the average power still at 23 dBm (a regulatory requirement for mobile UEs)

The Problem of 3.5 GHz in Rural Setting

Though LMLC can in theory work in the 3.5 GHz band, however, it will face practical problems. What are they? The most significant one is the poor propagation characteristics of this band compared with 700 MHz, i.e the radio waves in 3.5 GHz band will get attenuated much faster compared to the 700 MHz. Even LMLC is not likely to stretch the coverage to any level of comparison than what is possible in 700 MHz using conventional technology. This is apparent from the evaluation done by 5GIA for the 3.5 GHz band extracted under (Table 8, page 15). The parameters assumed are similar to those frozen for configuration c of LMLC.

One can clearly see that the CDF (cumulative distribution function) for PI/2 BPSK for ISD 12 Kms is stretched close to the y-axis. This means most of the users in the network will see a very low throughout. That translates to a spectral efficiency of "zero" at the cell edge. See under.

Note that the self-evaluation numbers used by TSDSI in their document (Section 20) are different as they have assumed a bandwidth of 60 MHz compared to 20 MHz used by 5GIA. Because of this reason, the configuration b (Non-LMLC) used in the M2412 document uses ISD of only 1.732 Kms.

Challenge of PI/2 BPSK in Lower Frequencies

PI/2 BPSK, the core technology to support LMLC, has challenges while one attempts to use it in the lower spectrum bands. The reasons are as under.

a) PI/2 BPSK works best in TDD spectrum bands and most of the lower spectrum bands are FDD. Since FDD is more efficient in lower spectrum bands than TDD.

b) PI/2 BPSK is not efficient for TDD bands with a limited amount of spectrum. Note that in 3.5 GHz band unless an operator chooses to acquire a full block of 100 MHz or more, PI/2 BPSK will not add much value in stretching the coverage, as the throughput at the edge of the network will become very poor.

The reason - using PI/2 BPSK one tries to boost the frequency components lying within the center frequency to a higher level (26 dBm) without amplifying the frequency components lying at the edge of the filter so that the average power still remains within the regulatory limit of 23 dBm. This is done to stretch the coverage from what would have been possible if one uses normal power of 23 dBm for all the frequency components lying within the bandwidth. But in doing so one has to compromise on the total throughput - especially at regions lying at the edge of the cell. Hence, unless one is empowered with sufficient bandwidth trade-off for throughput vs coverage for PI/2 BPSK is skewed. I.e throughput decreases exponentially in the urge to stretch the coverage. In higher spectrum bands bandwidth is not a constrain. But those at lower frequencies, bandwidth (per operator) will be a huge constrain. Hence, PI/2 BPSK is best used for TDD spectrum bands at higher frequencies and definitely NOT useful for 700 MHz even in TDD setting.

Evaluation of TSDSI's RIT

In the geneva meeting (WP 5D #34) held in Feb 2020, various groups evaluated TSDSI's RIT. All of them except TCOE found that TSDSI's performance claims in 700 MHz band are either par or less than compared to that of the 3GPP's. In the 3.5 GHz band, the comparison is apples to oranges, as TSDSI's self-evaluation has used totally different and relaxed assumptions compared to those used by the evaluation groups. Also, the 3.5 GHz band is not a candidate band for evaluation for LMLC and was decided at the beginning of this process, and therefore, India's self-evaluation of LMLC for this band was totally unnecessary. The summary is in the document embedded in the link. This resulted in the ITU meeting report which includes a statement suggested by India wherein it (India) acknowledges that some of its submission is outside the scope of the ITU evaluation process for IMT2020. See extracts as under (page 2).

"In conjunction with the above, TSDSI and India have provided the following statement for inclusion in this report:

The IEGs namely 5GIA, WWRF, AEG have provided evaluation results for the TSDSI RIT (IMT-2020/19(Rev.1)) through their submissions. TSDSI and India stated that TSDSI had brought before 34th meeting of WP 5D its response to the observations of the IEGs (as mentioned in section 11 in document IMT-2020/40, section 4 in document IMT-2020/51, and section 5 in document IMT 2020/41, respectively). However, these responses were not discussed in 34th meeting of WP 5D. WP 5D noted that the observations of the IEG’s and the corresponding responses of TSDSI were outside the scope of ITU evaluation on candidate IMT-2020 radio interface technologies."

Conclusion

From the above one can clearly conclude that TSDSI's self-evaluation of the configuration "c" (LMLC in 700 MHz) has passed the test of most of the evaluation groups on spectral efficiency numbers. Hence, theoretically, TSDSI's fork can qualify as an IMT 2020 5G standard. However, the question still remains whether such an outcome sufficient for India? The reason - a) PI/2 BPSK in 700 MHz would not add any significant value (to stretch the coverage), since 700 MHz India is an FDD band and cannot be used in a TDD setting as explained earlier; b) TSDSI's proposal for LMLC in the 3.5 GHz band is not an optimal band for rural and also could not get formally included in the WP 5D report since this band was not part of the original ITU documents (M2410, M2411, M2412) for LMLC; c) 3GPP has already developed technology which performs at par with that of TSDSI's; d) PI/2 BPSK technology (core technology for LMLC) is best used for higher spectrum bands (> 6 GHz), and not a suitable technology for 700 MHz. e) Due to (a),(b), (c) and (d), TSDSI will find it extremely challenging to convince 3GPP for the inclusion of its specifications into its core standards - absolutely essential to ensure interoperability, roaming, and economies of scale and productization of LMLC. Hence, the battle for India for the development and productization of its own 5G standards is far from over and has a long way to go.

(Views expressed are of my own and do not reflect that of my employer)

PS: Find the list of other relevant articles in the embedded link.