Spectrum Management for Dummies
Spectrum became a commodity of relevance with the invention of wireless communication. Most of the initial usage was anchored around defence applications. No special effort was made to harmonize spectrum across borders as the development of defence technologies was uncoordinated (on account of secrecy). Also, efficiency was least of the concerns, as plenty of it was available then. The advent of mobile telephony increased the importance of spectrum significantly. Two different versions of mobile technologies were developed in a gap of few years - the US AMPS (Advanced Mobile Phone System), and the European GSM (Global System for Mobile Communications), both using different spectrum bands. Since then the penetration of mobile phones has surpassed all expectations (McKinsey in 1980 predicted just 900K mobile subs in U.S by 2000) compelling us to focus on packing more users and optimizing device prices. Using spectrum efficiently allows us to pack more users per MHz, and aligning spectrum bands across countries reduces device prices by amortizing manufacturing and development costs across larger volumes (devices supporting many bands need more RF components). In spite of all the technological developments, spectrum continues to be scarce and its management complex - forcing mobile operators/developers to spend billions in auctions and equipment development cost (spectrum is 60% of CAPEX in India, and billions of dollars are spent in R&D worldwide). But on the other hand, WiFi devices are cheap and its spectrum available for free. Also, there are claims that some upcoming disruptive technologies can increase usage efficiency and mitigate the complexity of managing spectrum. The purpose of this note is to identify these opportunities/challenges and evaluate the ability of the disruptive technologies to overcome them.
Spectrum for Exclusive Use (Licensed)
Mobile network design has evolved from single tower covering the whole city talking to car mounted terminals, to thousands of mobile towers and easy to carry handheld terminals. The problem with single tower design was its inability to reuse spectrum and thereby significantly limiting the number of possible simultaneous users. Also, the mobiles had to transmit at very high power for the RF signals to reach the towers - making handheld terminals an impossibility. This led to the deployment of multiple towers whose coverage area is dependent upon mobile terminal's capability to transmit back to towers (mobile is constrained on antenna and battery capacity), and not that of the tower's capability to transmit to mobiles (towers can be designed to transmit at much higher power compared to mobiles). To increase network capacity the density of towers can be further increased but within a specified limit, as RF signals from towers too close to each other are likely to interfere (since they are transmitting at the same frequency).
For enabling ubiquitous coverage and reliable service, licensed spectrum has the best cost structure compared to its unlicensed counterpart. In the absence of uncoordinated outside interference (from other operator's networks), the power management (critical for spectral efficiency) can be network controlled - thereby making the handset design simpler and cheaper. Operator's ability to transmit at an optimal power reduces the number of towers and optical fiber terminal points - thus saving Capex and Opex. However, the biggest challenge is the high price of spectrum and the need to align spectrum bands with large a number of nations for amortizing device manufacturings costs over larger volumes. Fortunately, India has been able to assign licensed spectrum in the internationally harmonized band such as 800/900/1800/2100/2300 MHz.
Spectrum for Nonexclusive Use (Unlicensed)
Unlicensed spectrum does not guarantee exclusivity and therefore, there is no coordination between operators/users using this spectrum. Hence, other than managing interference from adjacent access points (the equivalent of towers), managing interference from users of other operator's network also becomes paramount. Physical isolation is created - a) by identifying unlicensed spectrum blocks at higher frequencies than lower, as at these frequencies the RF signals tend to attenuate faster compared to those at lower frequencies, b) by mandating the operators/users to transmit at much lower power compared to their licensed counterparts (4 watts ERP vs 20 watts ERP). This enables multiple numbers of networks to coexists in the same frequency band very close to each other. However, uncoordinated use makes it impossible to eliminate the possibility of interference and hence, the technologies using unlicensed spectrum uses frequency hopping techniques (similar to GSM) and dynamically choose a block of spectrum (among multiple) with least interference. Thus in order to enable reasonable data rates, an unlicensed operation need a much larger quantum of spectrum compared to its licensed counterpart (typically a block of 100/200 MHz) of which only a fraction (20/40 MHz) is used at a time. This is the reason why lower frequencies (less than 1 GHz) are not good for unlicensed operations as it is difficult to find such large blocks of spectrum at these frequencies. Also, RF signals (even at lower power levels) in the sub-GHz frequency tend to travel far causing severe interference to uncoordinated networks operating nearby.
It is very easy to set up an unlicensed network as it requires no regulatory clearance and spectrum is free. However, challenges are that it cannot be used for ubiquitous coverage and there is no guarantee on SLAs. The threat of interference on account of non-exclusive spectrum assignments forces one to over design the network with a large number of access points (example for a cricket stadium it could be as high as 500 to 700 APs!!!, same could be true for a busy railway station). This also significantly increases the number of backhaul access points compared to networks running on "exclusive" licensed spectrum. That is why in spite of the availability of cheaper WiFi devices, not much attempt has been made to roll out country-wide WiFi networks, but operators chose to pay billions of rupees to acquire licensed spectrum. India has assigned 2.4 GHz and 5 GHz band for unlicensed operations.
Spectrum for Nonexclusive Use (Licensed)
This licensing approach is used for assigning and managing spectrum for backhauling traffic from aggregation point to switches and visa verse. The same frequency spots can be assigned to different operators, but a geographical separation is maintained to prevent interference. This is possible by directing the antenna beams to a narrow solid angle. This serves two purpose a) increase in transmission gains, b) preventing leakage of unwanted power. Assignments are made in higher frequency bands where large blocks of spectrum are easy to find (for increased data throughput), and the attenuation at higher frequency band is managed by using towers of certain heights at both ends to bypass the ground clutter (all RF signals travel equally in free space). In India, licensed non-exclusive assignments are mostly made at 3.4 GHz band and above.
Spectrum for Exclusive Licensed Shared Access (LSA)
We know that spectrum licensed for "exclusive use" has the best cost structure and efficiency. However, to enable devices affordability these bands needs to be internationally harmonized to amortize the device's manufacturing and development costs over larger volumes. Unfortunately, some of the useful spectrum in these harmonized bands (900,1800,2100,2300,2500 etc) are occupied by state agencies (defence, space etc). Licensed Shared Access (LSA) is a means to unlock this valuable spectrum without disturbing the existing state's usages. This can be done by creating exclusion zones and databases marking state's existing operations. This database controls the base stations which in turn instructs the mobiles (using normal licensed spectrum) to lock into the LSA spectrum when and where it is available.
The basic advantage of this model is the ability of the LSA management system to direct and instruct the existing commercially available 3G/4G mobiles. This makes it fully backward compatible to all the existing devices, thereby ensuring economies of scale. LSA management system will be under the control of the state agencies who is the original user of the spectrum, with an override option of switching off commercial BTSs of operators (using the LSA spectrum) in case state agency needs to use the spectrum. This approach is different from cognitive radios which not only require special devices with intelligence (increasing cost) to sense network availability but also does not provide any control to the state agencies to restrict usage of its spectrum in the case of need- thus making them (state agencies) wary.
Licensed Shared Access will be used in European countries to open commercial spectrum band that is used by state agencies. It can be very useful in India, as a lot of commercial spectrum in mainstream bands is occupied by the defence and other PSU units, but very lightly used and lying fallow.
Spectrum for broadcast use (Licensed)
Broadcast services operate in the lower frequency bands of 470 to 648 MHz. Since the transmission is unidirectional (tower to radio receiver), the coverage only dependent on broadcast tower's ability to reach the user terminals. Hence, broadcast towers are much higher (100 meters vs 30 meters) and transmit at very high power (50KW vs 20W of mobile BTS) to cover hundreds of kilometers with a single broadcast tower. High power transmission makes adjacent frequency blocks unusable due to RF spillovers. These are whitespaces which are being targeted by special technologies like 802.11af and 802.22 to extend NOFN bandwidth to rural areas. However, the problem is on two accounts a) these technologies using whitespaces need to transmit at very low power to prevent interference to adjacent broadcast channels, and hence cannot leverage the better propagation characteristics of this low-frequency band; b) to extend 100 Mbps of NOFN bandwidth (NOFN is designed to carry 100 Mbps to villages) it will need at least 100 MHz (@ 1 Mbps/MHz) of continuous spectrum (not counting spectrum required for redundancy and managing interference).
Using power management strategies like cognitive radio and databases will further increase device complexities and therefore cost. Only advantages could be if the spectrum can be obtained free. But free spectrum will be difficult to justify as 80 to 90% of the spectrum in the block of 585 to 698 MHz is unused by broadcast services and therefore, are "not whitespaces" but "normal unused" spectrum, as in any other band.
Satellite Spectrum for Exclusive use (Licensed)
Satellite signals tend to spill over into neighboring countries and hence the spectrum assignments are controlled by international bodies like ITU. Local permissions are required only to set up in-country gateways. Thus, satellites also use licensed spectrum, as unlicensed low power transmissions will not travel far to cover the enormous distances involved (typically 100s of kilometers). The number of aggregation points being limited and hence a large amount of spectrum is required due to limited opportunities for reusing the same block of spectrum. Drones and balloons can improve this reuse factor as a large number of it can be used compared to satellites. Also, since they operate at much lower earth orbits (10 to 20 Kms), the risks of spillovers into neighboring countries is negligible, making the spectrum assignments for drone and balloons a national subject. Hence, new bands of "licensed" spectrum bands need to be carved out to enable its operations. Other options (like lasers) to ground access points can be tried, but the reliability could be an issue due to obstruction from hail and cloud cover.
Conclusions
India has assigned plenty of licensed spectrum which is currently not optimally used in rural areas. License spectrum can be used most efficiently and has the best cost structure and reach compared to that of the unlicensed. Providing broadband connectivity to 600 K villages is an ambitious goal. The task in hand is big and the potential benefits are huge, and hence all possible options should be tried to expedite it. Incentivizing operators with financial incentives (license and spectrum fee rebates) can act as a much-needed trigger to motivate them to extend broadband connectively to rural areas using matured and tested technologies in their existing licensed spectrum. This approach could be much faster and practically easy to implement than other untested options.
(The views expressed are my own and does not reflect that of my employer)
Thanks Debashish, the point that I was trying to point at that the approach that we are currently using for wireless has matured overtime, and network planning and use of spectrum (licensed, unlicensed, backhaul etc) all depends upon the services that one is trying to drive. If ubiquitous mobility is the objective, then best approach is licensed spectrum preferably terrestrial. If mobility is not the objective, and services needs are limited over small geography, then unlicensed spectrum could be a way to go, but at higher frequency as you need lots of it for frequency hopping. If backhaul is the objective, then you need to target bands where plenty of spectrum is available to support high data rates, as lower spectrum with smaller bandwidth will not be able to carry the required data rates, even if one transmits at normal power, as large blocks of spectrum is hard to find. Unless you choose the right spectrum for the application that you plan to drive, you will be always caught in the wrong foot as TVWS folks are today, as they plan to position a techology to provide a solution to a problem that does not exist. There are no wihitepace here. So what problem are you trying to solve through this special technology???
Experienced Professional in Sales, Marketing, Business Development , Corporate Affairs , Government interactions etc
9 年Hi Parag, My heartiest compliments for an extremely well researched article . I think after reading your article, I dare say no one should go beyond considering a trial or demonstration, when it comes to using a " special technology " and that too in an " unlicensed space or band " . Here, my indication is towards the much-touted TVWS technology which is trying to squeeze its way into the unlicensed White space of which there is none as the entire UHF band is lying empty or fallow . First of all, TVWS technology is being touted as a fiber extension or the " middle mile connectivity " from the Gram Panchayat to " somewhere " ( some learned people define it as a hamlet ! ) and says that the distance between the Gram Panchayat to this mysterious hamlet is more than 10-12 kms.!! Btw, if you look at the e-governance model from the Department of Panchayati Raj, there is nothing called a hamlet. It is District-Block-Gram Panchayat-Individual villages. In fact in the initial NOFN project, the government chose to lay the optical fiber upto the Gram Panchayat because the average distance between the GP & the villages is just under 3Kms.( with the exception of some areas like the North-East, some other hilly terrain and areas covered by dense forests .) and the expectation was that the connectivity to the last mile ( from the GP to the villages )would be done either directly using Fiber/Wireless ( 2G/3G/Wifi ) or through Satellite or Cable TV depending on the nature of the service to be provided and the type of service provider who wants to enter the area. If you go to the BBNL website and look at the GIS maps, you will find that the average distance between the Gram Panchayat & the village is around 2-3 kms and the average number of villages that a Gram Panchayat caters to is between 2-3 villages. Of course, if the NOFN fiber does not come up at the GP, then it needs to be backhauled from the Block to the GP and maybe beyond. In that case, we may encounter such mythical distances of 10-12 kms which have been touted as the reason If indeed the govt. chooses to go the wireless route for connecting the Block to the GP & beyond, instead of the fiber or chooses such to complement the fiber , then why should it choose TVWS ? First of all, TVWS which is the much-touted Rural BB technology provides 1Mbps/1Mhz spectrum consumed. If the objective is to provide 100Mbps at the GP as was the " National mandate " of the earlier NOFN project, then TVWS would need at least 100Mhz of spectrum ! Surely this cannot be a spectrum efficient broadband access/backhaul technology ! If the Govt decides to dilute that mandate to lower levels of say 20Mbps or so at the GP, then why do we need TVWS.? The existing licensed Microwave technologies can fulfill those requirements in a more efficient manner. Also, the whole TVWS discussions have been centred around 'best-effort 'basis instead of being able to provide five 9s reliability and guaranteed SLAs-interference prone band, low power signals, license exempt band. Surely,our rural people deserve a high quality and highly reliable broadband service and nothing less than the same grade of service being provided in urban areas ! Besides Net neutrality, India stands for Net Equality as well ! On the other hand, there are proven high capacity terrestrial wireless backhaul solutions which can drive capacities of upto 1-2Gbps at distances of around 10-13 Kms in higher licensed spectrum bands ! So, in my opinion , if the Govt is looking for a fiber-extension wireless technology for backhauling , it can choose other options. I am confident that the Govt of the day will consult all the stakeholders and will take the right decisions such as auctioning of the UHF spectrum to the licensed operators and leaving the choice of technology to the ones which are globally harmonised so as to bring down the cost of service to the rural areas and make it affordable .
Demystifying technology
9 年Lucid explaination
Government Affairs, Artificial Intelligence, Policy Advocacy, Program Management, Digital Transformation, Corporate Social Responsibility, and Intellectual Property
9 年you have explained a very difficult subject in a simple manner
This is very good. thanks. have shared