In the past decade India embarked and successfully completed “Aadhar”, the world’s largest social identification programme. This has led to fewer leakages in the subsidy scheme meant for the unprivileged.
The GoI has now embarked on the “Digital India”, a program to narrow down the Digital exclusivity and the Urban-Rural digital divide. It aims to provide connected to 500K villages and more mofussil towns. The advantages of such a program have been debated in multiple fora and the conclusion about its utility is forgone.
The back-bone of this program is a nationwide optical fibre and in some cases High throughput satellites ( HTS ), the radio technologies to provide the “last mile” and “meter” have been in question. Among the candidates, White spaces have been proposed as one of the preferred technologies. There have been debates on for its merits and demerits. A lot has been said by consultants but there has never been a deep dive. This is an attempt to dispel the myths from a practitioner that have been created over the past few years.
This article focuses on the technologies associated with the implementation of the Digital India program and the ‘myths’ that have been doing the rounds.
Myth 1: There is no white space in India
In India unlike in the US and Europe there are no private terrestrial broadcasters. The only broadcaster is the Government owned Doordarshan (DD). Since there is no analog switch off, for the moment the 470-646 MHz spectrum is vacant. Given this it might seem from a semantic perspective that there is no white space. But this in extremely narrow definition of White spaces.
White spaces are more than just empty channels between two active broadcast channels. White Spaces are a set of technologies that use cognitive and advanced signal processing to enable efficient and dynamic access of spectrum that incidentally uses the UHF/VHF spectrum. Given the ITU resolution WS devices are by definition secondary users and need to adhere to far stricter norms for spectral emissions. In India Xerox has become synonymous with the photocopier. Similarly white spaces encompass more than just spectral occupancy.
Myth 2 : White spaces are access technologies to connect phones or laptops to the fibre backhaul
As an engineer to me the distinction between access and backhaul spectrum sounds semantic as both are frequencies! However from an operator perspective access spectrum is an expensive resource that has to be utilized fully and efficiently. This makes sense. Given that the 700 MHz spectrum is priced quite high it is natural for operators to oppose “giving away” spectrum in the 470 MHz range for “free” (unlicensed).
As a semiconductor vendor supplying chipsets for white spaces, this should be music to my ears. However for a TVWS chipset to get into a cell phone, the cost structures must be that of Wi-Fi modules and to get to those economies of scale it would take quite a while. There has been talk implementing a tri-band Wi-Fi chipset that supports 802.11ac and 802.11af in a single chip. From an economics standpoint it won’t work as the margins for low end Wi-Fi chipsets are few 10s cents. Furthermore designing a small form-factor, high gain antenna at reasonable price points in the 470 MHz band is a huge challenge if not impossible.
Further cellular networks use a spectrum reuse of 3 to mitigate interference and reduce receiver costs. In an access network it makes sense to keep the receiver complexity low. However in back-haul networks which less sensitive to pricing interference mitigation techniques to improve overall throughput can be implemented. In the interim I believe that WS is suited to be a backhaul or “middle mile” technology that compliments the operator’s access spectrum especially for 2G/3G and Wi-fi small cells in rural areas.
Myth 3 : Spectral efficiency of white space technologies is poor
This is far from the truth and people raising these questions implicitly assume that Wi-Fi based CSMA/CD MAC protocols will be used for deploying White spaces. The more polite a protocol lower its spectral efficiency. Other MAC protocols such as OFDMA/TDM and standards like 802.22 are spectrally more efficient while retaining politeness to an extent. In fact some of these are better than LTE in the uplink. The table below gives a comparison of the various technologies used for white spaces.
|Cost||High BS, Low CPE||Low with standard antenna||Low Depending on volume|
|Actual Range||<3KM @ 40-70W BS power||<100mts @ 30dbm PA, 6dbi antenna||12-15Km @24dm PA, 11dbi antenna|
|Standard limitations for longer range||5-6 KM, Small and fixed GI, RACH issues with frequency correction||Very small GI’s||Variable and/or dynamic GI allowing up to 50KM|
|DL/UL speeds||Asymmetric, unsuited for UL heavy traffic like surveillance||Adaptable||Symmetric and adaptable to UL heavy traffic|
|Peak Spectral Efficiency downlink ( SISO )||4.1 b/s/hz long GI for longer distances||2.7 (a/g ), 3.6 (11n ) short GI||4.5 b/s/Hz for 1/32 GI|
|Peak Spectral Efficiency uplink (SISO )||2.5 b/s/Hz||2.7 (a/g ), 3.6 (11n ) short GI||4.5 b/s/Hz for 1/32 GI|
|Protocol Etiquette||None, doesn’t coexist with other transmitters||Extremely polite resulting in significant data throughput loss||Allows for fair share through a combination of database and co-existence procedures|
|Dynamic spectrum access||None, doesn’t coexist with other transmitters||No||Yes through a database and also spectrum ‘policing’|
A centrally coordinated data base that ensures “fairness” for competing users and operators ensures the best of fairness and spectral efficiency. Also in certain network architectures it becomes difficult to implement traditional MIMO due to the nature of the terrain. When pitched as a back-haul spectrum, there are other ways to improve spectral efficiency to about 15 bits/Hz which is far greater than the current LTE networks.
Myth 4: Channel bonding to increase data rates
Channel bonding is the easy way out to increase data throughputs especially when spectrally inefficient schemes are used. It works well for licensed spectrum and in the higher bands such as 5 and 6Hz where there are swathes of spectrum. In the UHF band spectrum is limited to about 120Mhz in India Therefore highly spectrally efficient schemes are imperative and traditional Wi-Fi and LTE type efficiency is not going to fly for large scale nationwide deployments. More so because of the spectrum reuse factor. Current cellular networks have a spectrum reuse of 3 which means that for every 1 Hz of spectrum 3Hz of spectrum is needed to actually deploy networks. The ideal is to get to a single frequency network. SFN networks usually put a heavy workload on the receivers and are therefore more likely to be implemented in a backhaul network than an access network
Myth 5 : Small Cell for urban, Macro cells for rural
There is no single network architecture that fits all. Where fibre POP’s are abundant it makes sense to use small cells as the reuse factor increases, throughputs increase and the overall cost of ownership is lowered.
Given that the fibre POP’s in the rural areas are limited and the range offered by macro cells it looks logical to use Macro cells. However cellular 3g/4g Macro cells consume about 1KW of power and are extremely expensive. This is one of the reasons operators are not able to service their service obligation that results in lots of LUT ( Low utilization towers) .
TVWS based backhauls change all of that. Pitching it as a backhaul alternative gives both cap-ex and ope-ex advantages. Operators can now provide 3g/4g services to rural areas.