Supportive 5G Infrastructure Policies are Essential for Universal 6G: Assessment using an Open-source Techno-economic Simulation Model utilizing Remote Sensing
VOLUME XX, 2017 Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.Doi Number
Supportive 5G infrastructure policies are essential for universal 6G: Evidence from an open-source techno-economic simulation model using remote sensing
Edward J. Oughton and Ashutosh Jha College of Science, George Mason University, Fairfax, VA 22030 USA S.P. Jain Institute of Management and Research (SPJIMR), Mumbai, 400058, India
Corresponding author: Edward J. Oughton (e-mail: eoughton [at] gmu [dot] edu).
This work was supported by George Mason University and S.P. Jain Institute of Management and Research.
ABSTRACT
Work has now begun on the sixth generation of cellular technologies (`6G`) and cost-efficient global broadband coverage is already becoming a key pillar. Indeed, we are still far from providing universal and affordable broadband connectivity, despite this being a key part of the Sustainable Development Goals (Target 9.c). Currently, both Mobile Network Operators and governments still lack independent analysis of the strategies that can help achieve this target with the cellular technologies available (4G and 5G). Therefore, this paper provides quantitative evidence which demonstrates how current 5G policies affect universal broadband, as well as drawing conclusions over how decisions made now for 4G and 5G affect future evolution to 6G. Using a method based on an open-source techno-economic codebase, combining remote sensing with least-cost network algorithms, performance analytics are provided for different 4G and 5G universal broadband strategies. As an example, the assessment approach is applied to India, the world`s second-largest mobile market and a country with notoriously high spectrum prices. The results demonstrate the trade-offs between technological decisions. This includes demonstrating how important current infrastructure policy is, particularly given fiber backhaul will be essential for delivering 6G quality of service. We find that by eliminating the spectrum licensing costs, 100% 5G population coverage can viably be achieved using fiber backhaul. In conclusion, supportive 5G infrastructure policies are essential in providing a superior foundation for evolution to 6G.
INDEX TERMS
Broadband, 5G, 6G, economic, techno-economic, open-source.
I. INTRODUCTION
A flurry of engineering research on 6G is now underway [1]–[15]. Already the provision of global broadband coverage to both unconnected and poorly connected users has been a central development area [16]–[23]. This topic received less attention than preferred in the previous 5G R&D standardization process. Broadband connectivity is becoming increasingly important to ensure sustainable economic development. There is a particular focus on reducing the digital divide in low- and middle-income countries to support the delivery of the United Nation’s Sustainable Development Goals. The global coronavirus pandemic has only increased the political impetus for broadband deployment because it makes digital connectivity even more essential [24], [25]. One of the most cost-effective approaches for delivering broadband over wide geographic areas is via cellular technologies, particularly using 4G, but in the future, this may include 5G too. These cellular technologies are efficient at moving large quantities of data, thus lowering the delivery cost per bit. However, rural connectivity has generally been an afterthought in cellular standardization, meaning the business case for deployment is often weak [26], [27]. Many 6G papers are focusing mainly on urban scenarios, which would lead this generation into the same issues as 5G [28]. Indeed, questions are being asked if 6G needs to play more of a role [29], whether by new technologies or spectrum management innovation [30]–[37]. Therefore, an emerging aim for 6G is to achieve a dramatic price reduction in cost
VOLUME XX, 2021 compared to previous technologies [38]–[40]. Our conjecture is that 5G focused too much on providing higher capacity, but not enough on reducing cost and providing affordable broadband for the unconnected. Even with the technologies standardized, the engineering community as well as Mobile Network Operators (MNOs) and governments, still lack effective open-source analytics to help them understand the investment strategies for universal broadband, particularly how these strategies play out in spatio-temporal terms (which is almost always overlooked in both 5G and 6G research) [41], [42]. This provides strong motivation for this paper's content, which aims to consider both the technologies we have available for deployment now (4G and 5G), but approach their evaluation with consideration for a post-5G world, particularly given the emerging research on 6G technologies. Although the deployment of 6G is still many years away, numerous high-level 6G positioning papers have been published focusing on the qualitative theoretical discussion of ‘what should 6G be?’ [43]–[55]. We believe we need to start considering the long-term evolution of current technologies to 6G now, but with a greater quantitative focus on cost-effectiveness (with this paper being a demonstrable example). Despite the grand policy goals for the next decade, we are left with many engineering and economic questions regarding broadband deployment in unconnected areas. When will 4G or 5G reach unconnected users? How will decisions we make now prevent further transition to 6G when even more ambitious capacity and latency requirements are expected? With these issues in mind, the following research questions are identified: 1. How do different 4G and 5G strategies quantitatively perform in viably delivering universal broadband coverage? 2.
What impact do spectrum price changes have on coverage-focused universal broadband strategies? 3.
Can conclusions be developed to inform current 5G policies and future 6G standardization and deployment? To answer these research questions, the remainder of this paper is structured as follows. The next two sections provide an overview of the related literature, followed by an articulation of the generalizable research method in Section IV. The application of the method is presented in Section V, with the results reported in Section VI. Finally, a discussion is undertaken in Section VII which addresses the first two research questions, based on the results obtained. Finally, the third research question is answered in Section VIII as relevant conclusions are identified.
II. WHY 5G POLICY MATTERS TO ENGINEERS
In recent years 5G has become wrapped up in an international competition between nations, for example between the USA, China, South-Korea, Japan, the UK and Europe [56]–[63]. There has been a focus on new technological opportunities to provide enhanced capacity and coverage [64]–[77], as well as cybersecurity issues [78]–[82], particularly how these may, directly and indirectly, affect industrial sectors [83], [84]. However, deploying advanced 5G technologies is hitting various economic roadblocks. Firstly, the Average Revenue Per User (ARPU) in mobile markets has either remained static or been in decline in most countries, falling by approximately 1% annually [85]. This is troubling for MNOs who are likely to experience little in the way of new revenue from 5G but are simultaneously being pressured by governments to make large infrastructure investments which deliver on the three main use cases of Enhanced Mobile Broadband (eMBB), Ultra Reliable Low Latency Communication (uRLLC) and Massive Machine Type Communication (mMTC) [86]. Secondly, the 5G regulatory burden being placed on MNOs is considerable, with significant resources needing to be allocated to purchasing spectrum licenses, which require substantial capital investments [87]. Indeed, there are fiscal ramifications arising where high spectrum prices leave little available capital for expansion to less viable areas [88]. These issues do not bode well for deploying 5G to less attractive regions, which could reinforce the digital divide. Recent literature concerning the deployment of 5G has mainly focused on the policy and economic implications for leading high-income economies, with only a few examples considering the implications for low- and middle-income countries where most unconnected users reside [89], [90]. Even in leading economies, the policy landscape is still evolving to work out how best to help deliver the potential benefits of 5G, particularly given the embryonic deployment of these technologies. But what has not changed is the desire to extract the maximum amount of spectrum revenue from the sale of new licenses, which to a certain extent is at odds with the policy desire of providing ubiquitous high capacity broadband connectivity to help spur the digital economy. In summary, there needs to be a much greater quantitative focus on how we will deliver universal broadband at a practical level, including quantification of the ramifications of national policy decisions, for example, on spectrum pricing.
III. DELIVERING UNIVERSAL BROADBAND
Universal service is a policy that aims to provide all households and businesses with access to a given utility, such as broadband, electricity, or water [91], [92], to be able to reduce access inequality [93], [94]. One of the oldest examples includes universal access to fixed telephone services, which have existed for almost a century [95], [96]. Still, as demand for legacy services has declined, requirements have been adapted to keep up with the digital economy's growth and demand [97]. New universal service policies have also been frequently introduced, particularly when a single previously nationalized service provider is
VOLUME XX, 2021 privatized and opened to market forces [98]–[101]. In such a case, the policy aim is to ensure that users in areas of market failure, where the cost of supply exceeds the amount that users are willing to pay, do not undergo a loss of service, while simultaneously taking advantage of the benefits of competitive markets in viable areas [102]. Depending on the historical evolution of a telecom market, this can differ by country [103], with some instead favoring the reduction of prices for underserved households [104], [105]. More recently, universal service requirements have been applied to mobile broadband markets via new spectrum licensing regimes. This has enabled the delivery cost to be subjected to market efficiencies via the auction bidding process [106], simultaneously delivering on equity and efficiency objectives [107]. Different designs have been implemented in many countries, each reflecting heterogenous institutional preferences, such as the degree of market involvement and the level of top-down government control [108]–[110]. There are mixed results, however. Although universal broadband aims are admirable, many people are still not connected to a decent service, indicating mixed success in achieving broadband policy objectives. IV. OPEN-SOURCE TECHNO-ECONOMIC ASSESSMENT
A generalizable model is now presented, which enables the techno-economic assessment of universal broadband strategies using either 4G or 5G (but could also be adapted in the future to evaluate candidate 6G technology strategies). An open-source software codebase is developed which enables application to any country in the world. The assessment utilizes both simulation techniques and a scenario approach to provide the ability to ask ‘what if’ questions, which is a common methodological approach for infrastructure assessment [111]–[115], as applied here to a ‘hypothetical MNO’. The aim is to use average information to broadly represent market share, spectrum portfolio, and sunk investments in assets to provide a general understanding of different strategic decisions on cellular technologies. The generalizable assessment method is visualized in Figure 1. A set of scenarios can be used to explore different user capacities. The targets are segmented based on urban, suburban, or rural settlements, reflecting the fact that the engineering requirements and thus the economic costs of delivery are significantly different between these areas. Current universal broadband targets being used by the UN Broadband Commission range from 2 Mbps (enabling most web browsing and email activities) up to 10 Mbps (enabling HD multimedia). In terms of strategies, there are a wide variety of technologies available for MNOs. Firstly, cellular technologies have proven to be cost-effective in providing wide-area connectivity [38], particularly in areas with no existing fixed broadband infrastructure to upgrade. Either 4G or 5G technologies are the main options currently being considered for broadband connectivity. Secondly, while there are significant choices to make in terms of the RAN technology, the backhaul connection type is also an important consideration in order to provide a cost-effective link from the cell tower to the nearest fiber Point of Presence (PoP) [116]. In many countries, wireless backhaul is still the dominant technology because the costs of deployment are lower than other options. The number of local users for different data services must be estimated, which is a function of the local population, the number of cell phone users, and the types of cell phone users. To obtain the total addressable market for cellular services in the 𝑖 th area, the population is required ( 𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑖 ). Using the 1 km WorldPop population dataset, derived from global satellite imagery, it is possible to extract an estimation of the local population for any location in the world [117], [118]. Penetration data is then required on the total number of wireless subscribers, providing the cellular penetration rate for the 𝑖 th area ( 𝐶𝑒𝑙𝑙𝑃𝑒𝑛 𝑖 ). Additionally, a smartphone penetration rate is required ( 𝑆𝑃𝑃𝑒𝑛 ). Thirdly, the hypothetical MNO only carries traffic for its subscribers. Hence, users are segregated across the available networks uniformly, by dividing the user base by the number of networks in operation (
𝑁𝑒𝑡𝑤𝑜𝑟𝑘𝑠 ). As we aim to deliver 4G and 5G services to smartphone users (as users need this type of device to access them), we thus estimate the number of smartphone users (
𝑆𝑃𝑈𝑠𝑒𝑟𝑠 𝑖 ) in the 𝑖 th area as in eq. 1. 𝑆𝑃𝑈𝑠𝑒𝑟𝑠 𝑖 = 𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑖 ∙ (𝐶𝑒𝑙𝑙𝑃𝑒𝑛 𝑖
100 ) ∙ (𝑆𝑃𝑃𝑒𝑛100 )𝑁𝑒𝑡𝑤𝑜𝑟𝑘𝑠 (1) The revenue generated locally (
𝑅𝑒𝑣𝑒𝑛𝑢𝑒 𝑖 ) can also be estimated in an area by allocating consumption user tiers to local users using nightlight luminosity. Using nightlight luminosity remotely-sensed via satellites is an established way to remotely differentiate geographic regions based on the estimated level of development [119]. Hence, this approach can be used to estimate the Average Revenue Per User ( 𝐴𝑅𝑃𝑈 𝑐 ) for cellular users, broadly segmenting areas with low luminosity into lower ARPU categories and higher luminosity into higher ARPU categories. The logic is based on areas with higher socioeconomic status being able to afford to spend more on consuming electricity, which is therefore correlated with being able to spend more on cellular services. VOLUME XX, 2021 Figure 1 Structure of modeling approach
Using the NOAA DMSP-OLS global nightlight layer, luminosity levels are allocated a ‘Digital Number’ (DN) ranging from 0 to 64 (from no luminosity to very high luminosity) [120]. We allocate areas above 3 DN into the higher ARPU category, areas below 1 DN into the lowest APRU category, and areas falling between into the middle ARPU category. In eq. (2), we then convert these estimates into the revenue per area (km ) given consumption of smartphone ( 𝑆𝑃𝑈𝑠𝑒𝑟𝑠 𝑖 ) and regular cell phone users (C 𝑒𝑙𝑙𝑈𝑠𝑒𝑟𝑠 𝑖 ). 𝑅𝑒𝑣𝑒𝑛𝑢𝑒 𝑖 = (𝑆𝑃𝑈𝑠𝑒𝑟𝑠 𝑖 ∙ 𝐴𝑅𝑃𝑈 𝑐 ) + (𝐶𝑒𝑙𝑙𝑈𝑠𝑒𝑟𝑠 𝑖 ∙ 𝐴𝑅𝑃𝑈 𝑐 )𝐴𝑟𝑒𝑎 𝑖 (2) Future revenue needs to be discounted to the Net Present Value (NPV) over the assessment period to account for the time value of money, using a discount rate of 5%. There also needs to be an estimate of the quantity of user data generated to design a network to transport traffic. The estimated level of data traffic (
𝑇𝑟𝑎𝑓𝑓𝑖𝑐 𝑖 ) in each area (km ) is calculated for the given number of smartphone users ( 𝑆𝑃𝑈𝑠𝑒𝑟𝑠 𝑖 ) and the scenario defined capacity target for different urban, suburban or rural settlement patterns ( 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦𝑇𝑎𝑟𝑔𝑒𝑡 𝑠 ) using eq. (3). 𝑇𝑟𝑎𝑓𝑓𝑖𝑐 𝑖 = (𝑆𝑃𝑈𝑠𝑒𝑟𝑠 𝑖 ∙ 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦𝑇𝑎𝑟𝑔𝑒𝑡 𝑠 )𝐴𝑟𝑒𝑎 𝑖 (3) VOLUME XX, 2021 Often a geolocated site dataset is not available, only estimates of towers by region, requiring a disaggregation to be carried out (see [121]–[125] for tower counts by country). Therefore for each statistical unit, data are required for the total population (
𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 ), the total number of sites (
𝑇𝑜𝑤𝑒𝑟𝑠 ), and the percentage population coverage (
𝐶𝑜𝑣𝑒𝑟𝑎𝑔𝑒 ). To obtain the number of towers (
𝑇𝑜𝑤𝑒𝑟𝑠 𝑖 ) in the 𝑖 th local area, the method reported in eq. (4) allows us to estimate using the area population ( 𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑖 ). All areas initially need to be sorted using population density, to allocate towers to the most densely populated areas first, as any rational infrastructure operator would act. Once all towers have been allocated, the remaining areas without coverage have no towers, reflecting areas of market failure and thus no existing connectivity. 𝑇𝑜𝑤𝑒𝑟𝑠 𝑖 = 𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑖 ∙ 𝑇𝑜𝑤𝑒𝑟𝑠(𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 ∙ (𝐶𝑜𝑣𝑒𝑟𝑎𝑔𝑒100 ) (4) The technology type per site is allocated by taking global coverage polygons from the Mobile World Coverage Explorer. The disaggregated site estimates undertaken using eq. (4) are then intersected with these polygons to obtain the estimated site technology, such as 2G, 3G or 4G. At the backhaul technology type for each cell site is not available, we utilize data on the composition of technologies for macro cell sites by region [126], which for example, is approximately 1% fiber, 3% copper, 94% wireless microwave and 2% satellite in South Asia. As we do not have spatial data to estimate backhaul type, a sequential probability can be applied, which allocates the percentage of fiber to sites in the densest local areas and the percentage of satellite to the sites in the least dense areas. Copper and microwave are allocated proportionally to the percentage of sites in the middle of the distribution. Importantly, the backhaul composition allocated in this way ensures aggregated estimates match the data source, avoiding additional modeling uncertainty. Network maps for telecom operators are digitized and used to establish existing sunk investments in fiber. The structure derived is treated as the network edges and then used to estimate the network nodes. Without data to inform the existing nodes, an estimate is necessary. Hence, a settlement layer is developed where 1 km cells above a set threshold are extracted from the raster layer, with spatially proximate cells being summed and those exceeding a specific settlement size being added to the agglomeration layer. Fiber connectivity is then treated as existing at any town over 10,000 inhabitants within 2 kilometers of a core edge, as a rational infrastructure operator would want to maximize the sale of connectivity services to support the building of a long-distance fiber network. We then also connect any regions without a core node, using a least-cost design. The largest regional settlement is connected to the closest existing core node with a new fiber link. Finally, regional fiber networks are deployed, which connect settlements over 10,000 total inhabitants into an existing core node by building a new fiber link. The least-cost fiber network design consists of a minimum spanning tree estimated using Dijkstra's algorithm. The least-cost RAN design consists of three main stages, including using a 3GPP 5G propagation model to obtain the spectral efficiency [127], estimating the total channel capacity per spectrum band given a spectrum portfolio, and then finding the least-cost backhaul structure to connect new cell sites into existing fiber Points of Presence (PoPs). Firstly, there are three main ways to enhance the capacity of a wireless network, such as increasing the spectral efficiency of the technology in use, adding new spectrum bandwidth, or increasing the spectral reuse by building new cell sites. A generalizable system model is used to estimate the capacity of a cellular network based on using a stochastic geometry approach. The mean Network Spectral Efficiency ( 𝜂̅ 𝑎𝑟𝑒𝑎𝑓 ) (bps/Hz/km ) for an area is estimated using the average number of cells per site ( 𝜂̅ 𝑐𝑒𝑙𝑙𝑠 ) and the density of co-channel sites ( 𝜌 𝑠𝑖𝑡𝑒𝑠 ) utilizing the same carrier frequency ( 𝑓 ), as defined in eq. (5). 𝜂̅ 𝑎𝑟𝑒𝑎𝑓 = 𝜂̅ 𝑐𝑒𝑙𝑙𝑠𝑓 ∙ 𝜌 𝑠𝑖𝑡𝑒𝑠 (5) Hence, for all frequencies in use, the capacity of the total area (
𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑎𝑟𝑒𝑎 ) is estimated via the multiplication of the Network Spectral Efficiency ( 𝜂̅ 𝑎𝑟𝑒𝑎𝑓 ) by the bandwidth of the carrier frequencies ( 𝐵𝑊 𝑓 ) in use, as in eq. (6). 𝐶𝑎𝑝𝑎𝑐𝑖𝑡𝑦 𝑎𝑟𝑒𝑎 = ∑ 𝜂̅ 𝑎𝑟𝑒𝑎𝑓𝑓 𝐵𝑊 𝑓 (6) Lookup tables are generated via a simulation process using stochastic geometry via the open-source Python Simulator for Integrated Modelling of 5G. The 3GPP 5G propagation model is used to estimate the Signal-to-Inference-plus-Noise-Ratio (SINR), followed by obtaining the spectral efficiency via modulation and coding lookup tables [127]–[129]. For full details of the radio capacity modeling method, including all simulation parameter values, see [130]. To estimate the Quality of Service, the capacity provided for the cell edge rate (Mbps per km ) is mapped to a particular environment (e.g., urban or rural), antenna type VOLUME XX, 2021 (e.g., 2x2 or 4x4 MIMO), carrier frequency, cellular generation, and desired confidence interval, ready for querying during the modeling process. Initially, using a defined spectrum portfolio, a baseline capacity can be estimated for the current level of infrastructure availability. Then during the modeling process, the same approach can be used to estimate the number of required sites to meet different scenarios of capacity per user. Finally, the backhaul cost to either connect newly deployed cell sites or upgrade the link on existing sites is defined based on the technology strategy being tested and the mean path distance. By accounting for the density of the existing fiber PoPs ( 𝑑𝑒𝑛𝑠𝑖𝑡𝑦 𝑖 ) in the 𝑖 th region, the mean path distance ( 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑖 ) can be estimated as per eq. (7). 𝑑𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑖 = √ 1𝑑𝑒𝑛𝑠𝑖𝑡𝑦 / 2 (7) The estimated distance can then be converted to the required fiber investment given the cost per kilometer. For the wireless backhaul, the required investment is also segmented depending on the required distance and the size of the equipment needed. Links under 15 km use a set of small backhaul units, and links over 30 km use a set of large backhaul units, whereas those in between use the medium-sized variant. Once a least-cost network has been designed for a particular scenario and strategy, any new greenfield assets or brownfield infrastructure upgrades need to be costed. As there is a time dimension to the assessment over the study period, all costs are discounted using a 5% discount rate to produce the NPV to the current initial period. The network architecture illustrated in Figure 2 is used to upgrade legacy cellular sites to either of the chosen technologies using the unit cost information reported in Table 1. A literature review is used to evaluate the yielded cost estimates against other cellular deployments for typical three-sector macro cells. The greenfield estimates broadly match an equipment cost of $32k, a site build cost of $20k, and an installation cost of $5k [59], [131]–[138]. Any backhaul or core network upgrades are explicitly modeled based on the distances needing to connect the assets. An annual operational cost is treated as 10% of the capital expenditure, as in prior literature [139]. The cost estimates here do not yet include all the additional administration costs an MNO has to bear, which are added later. For example, these estimates are below the site costs used in other studies, ranging from $100-200k each. Moreover, as the capital needs to be borrowed via money markets, a Weighted Average Cost of Capital (WACC) is applied, reflecting lending risk. Spectrum prices can be developed by taking recent auction results via any available global spectrum database and breaking down each frequency into the US dollar cost per Hertz per member of the population ($/Hz/pop). Such an approach accounts for differences in bandwidth and country population size, which can cause large differences in aggregate spectrum values. Sub-1 GHz bands are treated as ‘coverage’ spectrum and usually have higher prices due to favorable propagation characteristics. In contrast, frequencies over 1 GHz are treated as ‘capacity’ spectrum and usually have lower costs due to less favorable propagation characteristics. Asset Cost ($USD)
Sector antenna 1,500 Remote radio unit 3,500 IO fronthaul 1,500 Processing 1,500 IO S1-X2 1,500 Control unit 2,000 Cooling fans 250 Power supply 250 Battery power system 10,000 Base Band Unit Cabinet 200 Tower 5,000 Civil materials 5,000 Transportation 5,000 Installation (per site) 5,000 Site rental (urban) (per site) 15,000 Site rental (suburban) (per site) 5,000 Site rental (rural) (per site) 1,000 Router 2,000 Backhaul - wireless (small) 20,000 Backhaul - wireless (medium) 30,000 Backhaul - wireless (large) 60,000 Backhaul - fiber (urban) (m) 20 Backhaul - fiber (suburban) (m) 10 Backhaul - fiber (rural) (m) 5 Regional fiber link (m) 2 Regional fiber node 100,000 Core fiber link (m) 4 Core fiber node 50,000 Table 1 Unit costs
VOLUME XX, 2021 Figure 2
Network architecture for cellular upgrades to 4G and 5G Once all these components are combined, the 𝑖 th area operator cost ( 𝑃𝑟𝑖𝑣𝑎𝑡𝑒_𝐶𝑜𝑠𝑡 𝑖 ) is comprised of the investment in the network ( 𝑁𝑒𝑡𝑤𝑜𝑟𝑘 𝑖 ), any administration ( 𝐴𝑑𝑚𝑖𝑛𝑖𝑠𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑖 ), prevailing spectrum prices ( 𝑆𝑝𝑒𝑐𝑡𝑟𝑢𝑚 𝑖 ), necessary corporation tax ( 𝑇𝑎𝑥 𝑖 ), and a fair profit margin ( 𝑃𝑟𝑜𝑓𝑖𝑡 𝑖 ), as illustrated below in eq. (8): 𝑃𝑟𝑖𝑣𝑎𝑡𝑒_𝐶𝑜𝑠𝑡 𝑖 = 𝑁𝑒𝑡𝑤𝑜𝑟𝑘 𝑖 + 𝐴𝑑𝑚𝑖𝑛𝑖𝑠𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑖 + 𝑆𝑝𝑒𝑐𝑡𝑟𝑢𝑚 𝑖 + 𝑇𝑎𝑥 𝑖 + 𝑃𝑟𝑜𝑓𝑖𝑡 𝑖 (8) To obtain the components of eq. (8), we need to estimate the structure for the network cost, spectrum, taxation and profit. By taking the sum of the Radio Access Network ( 𝑅𝐴𝑁 𝑖 ), backhaul ( 𝐵𝑎𝑐𝑘ℎ𝑎𝑢𝑙 𝑖 ) and core ( 𝐶𝑜𝑟𝑒 𝑖 ) in the 𝑖 th area the Network cost ( 𝑁𝑒𝑡𝑤𝑜𝑟𝑘 𝑖 ) can be obtained following eq. (9): 𝑁𝑒𝑡𝑤𝑜𝑟𝑘 𝑖 = 𝑅𝐴𝑁 𝑖 + 𝐵𝑎𝑐𝑘ℎ𝑎𝑢𝑙 𝑖 + 𝐶𝑜𝑟𝑒 𝑖 (9) The administration cost ( 𝐴𝑑𝑚𝑖𝑛𝑖𝑠𝑡𝑟𝑎𝑡𝑖𝑜𝑛 𝑖 ) is treated as a percentage of the network and represents the large amount of money that needs to be spent on running an MNO, including on real estate, salaries, vehicle fleets, R&D, etc. This can be up to 30% in high-income economies [140]. Next, to obtain the spectrum cost ( 𝑆𝑝𝑒𝑐𝑡𝑟𝑢𝑚 𝑖 ) we need to take each of the 𝑓 frequencies in the 𝑖 th area and multiply the dollar value per MHz per capita ( 𝐶𝑜𝑠𝑡_$_𝑀𝐻𝑧_𝑝𝑜𝑝 𝑓 ), channel bandwidth ( 𝐵𝑎𝑛𝑑𝑤𝑖𝑑𝑡ℎ 𝑓 ) and population ( 𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑖 ), as per eq. (10): 𝑆𝑝𝑒𝑐𝑡𝑟𝑢𝑚 𝑖 = ∑ 𝐶𝑜𝑠𝑡_$_𝑀𝐻𝑧_𝑝𝑜𝑝 𝑓𝑓 ∗ 𝐵𝑎𝑛𝑑𝑤𝑖𝑑𝑡ℎ 𝑓 ∗ 𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛 𝑖 (10) For the taxation rate ( 𝑇𝑎𝑥_𝑅𝑎𝑡𝑒 ) in the 𝑖 th area, the total tax due ( 𝑇𝑎𝑥 𝑖 ) can be calculated given the sum of the network cost ( 𝑁𝑒𝑡𝑤𝑜𝑟𝑘 𝑖 ) and spectrum cost ( 𝑆𝑝𝑒𝑐𝑡𝑟𝑢𝑚 𝑖 ), as detailed in eq. (11): VOLUME XX, 2021 𝑇𝑎𝑥 𝑖 = 𝑁𝑒𝑡𝑤𝑜𝑟𝑘 𝑖 ∗ ( 𝑇𝑎𝑥_𝑅𝑎𝑡𝑒100 ) (11) As an MNO takes a risk in a private market, there needs to be a fair return for any 4G or 5G infrastructure provision. Therefore, in the 𝑖 th area, a profit margin ( 𝑃𝑟𝑜𝑓𝑖𝑡 𝑖 ) is added for all investments (in addition to the WACC risk premium), as stated in Eq. (12): 𝑃𝑟𝑜𝑓𝑖𝑡 𝑖 = ( 𝑁𝑒𝑡𝑤𝑜𝑟𝑘 𝑖 + 𝑆𝑝𝑒𝑐𝑡𝑟𝑢𝑚 𝑖 + 𝑇𝑎𝑥 𝑖 ) ∗( 𝑃𝑟𝑜𝑓𝑖𝑡_𝑀𝑎𝑟𝑔𝑖𝑛100 ) (12) An important part of the model is that excess profits (>10%) are reallocated via user cross-subsidization to unviable locations to explore how the total revenue in the market could encourage infrastructure rollout in harder-to-reach places. Without such a mechanism, the only viable areas would be dense urban and suburban areas, and therefore there would not be any further upgrade to other locations (which does not necessarily match reality). After accounting for any reallocated capital via user cross-subsidization, any shortfall in connecting unviable locations would consequently require a state subsidy. V. APPLICATION
An assessment period of 2020-2030 is used to capture cellular deployment over the next decade focusing on testing either 4G or 5G Non-Standalone (NSA) strategies. India is used as an example as the country fits with the key trends already identified as affecting the deployment of 5G. Firstly, India’s ARPU has been on a constant decline in recent years, resulting in plummeting revenues for various incumbent MNOs [141]. Amidst such a scenario, there are widespread apprehensions concerning the financial feasibility of deploying 5G networks and provisioning 5G services in the country. Secondly, India is regarded as having some of the highest spectrum prices globally, which raises issues around how aggressive the reserve price may be for 5G bands. Additionally, India has a well-known issue with cellular backhaul availability [142], [143]. India is divided into twenty-two telecom zones, referred to as telecom circles, with each having a different potential for financial returns and, therefore, different spectrum prices. This creates a considerable administrative burden on an MNO to obtain licenses. In India, researchers have already been evaluating the feasibility of 5G deployment across a wide range of technologies [144]–[157]. With the existing level of capacity between 3-10 Mbps per user, there is considerable scope for improvement, although we should recognize that these estimates are based on crowdsourced data from mainly urban users, so the situation is likely much worse in rural areas [158].
Region Name Code ARPU Tier ($) Spectrum cost (<1 GHz) ($/Hz/ pop) Spectrum cost (>1 GHz) ($/Hz/ pop) Low Medium High Andhra Pradesh AP $0.6 $1.2 $1.9 $2.22 $0.54
Assam AS $0.5 $1.0 $1.6 $0.70 $0.13
Bihar BR $0.4 $0.8 $1.2 $0.19 $0.05
Delhi DL $0.5 $1.0 $1.5 $10.18 $3.04
Gujarat GJ $0.5 $1 $1.6 $1.11 $0.32
Haryana HP $2 $3 $6 $0.89 $0.25
Himachal Pradesh HR $0.4 $0.8 $1.1 $0.67 $0.28
Jammu & Kashmir JK $0.5 $1 $1.5 $0.59 $0.13
Karna-taka KA $0.6 $1.2 $1.8 $1.19 $0.46
Kerala KL $0.6 $1.3 $1.9 $1.2 $0.38
Kolkata KO $0.4 $0.9 $1.3 $11.76 $3.09
Madhya Pradesh MH $0.6 $1.1 $1.7 $1.27 $0.29
Mahara-shtra MP $0.5 $1.0 $1.4 $0.71 $0.13
Mumbai MU $0.6 $1.2 $1.9 $7.39 $2.29
North East NE $0.5 $1.1 $1.6 $0.50 $0.09
Orissa OR $0.5 $0.9 $1.4 $0.34 $0.08
Punjab PB $0.5 $1.1 $1.6 $1.07 $0.42
Rajast-han RJ $0.5 $1.1 $0.5 $0.58 $0.23
Tamil Nadu TN $0.6 $1.3 $1.9 $1.22 $0.89
UP (East) UE $0.4 $0.7 $1.1 $0.24 $0.01
UP (West) UW $0.4 $0.8 $1.2 $3.92 $1.43
West Bengal WB $0.5 $0.9 $1.4 $0.21 $0.05
Table 2 ARPU consumption tiers Scenario 1 focuses on a basic set of targets for urban, suburban and rural areas consisting of 25, 10 and 2 Mbps, respectively. Secondly, in Scenario 2, an intermediate set of targets for urban, suburban and rural areas focus on delivering 50, 20 and 5 Mbps, respectively. Finally, in Scenario 3, a highly ambitious set of capacities for urban, suburban and rural areas aim to deliver 100, 30 and 10 Mbps, respectively. The scenarios selected represent a broad range of options to provide insight into how the delivered capacity affects cost, and therefore the deployment of universal broadband using either 4G or 5G across different urban-rural settlement patterns in India.
VOLUME XX, 2021 Figure 3 Subscriber and smartphone forecasts
VOLUME XX, 2021 Figure 4 Demand and supply density metrics for the year 2020
Figure 5: Scenario viability by technology strategy The telecom circles are listed by name and abbreviation code in Table 2, along with the ARPU consumption tiers per user in each area. The demand forecasts developed can be viewed in Figure 3 for all regions assessed. The forecasts visualize both the number of unique mobile subscribers and the adoption of smartphones. For the cellular penetration rate, the number of unique subscribers is obtained from the historical data (2010-2020) and used for forecasting over the study period to 2030 [159]. Historical data is not available for smartphone penetration; therefore, a set of consistent growth rates are used to forecast smartphone penetration across both urban and rural regions. In Figure 4, both the demand and supply metrics are presented nationally by decile for India, for both the total market and a single modeled MNO with 25% market share. In developing the settlement layer, most telecom circles use a cell threshold of 500 inhabitants km with a settlement threshold of 1000 total inhabitants. The exceptions include Mumbai, Delhi, and Himachal Pradesh, which use a cell threshold of 1000 inhabitants km and a settlement threshold of 20,000 total inhabitants. The resulting points layer of all settlements is used to develop the least-cost network routing structure. To incorporate both the existing as well as the planned fiber network across the VOLUME XX, 2021 settlements, the geospatial data for the Indian railway network is used, since fiber deployments are laid along the railway lines [160]. If settlements are within a 5 km buffer of the railway line they are treated as having fiber connectivity because the rational aim of deploying the network is to maximize access to as many settlements as possible. An average MNO spectrum portfolio for India is identified, which includes deploying 4G in Frequency Division Duplexing (FDD) using 850 MHz (MIMO 2x2) with either 2x1.25 or 2x2.25 MHz of bandwidth for paired channels. Additionally, 1800 MHz is available with 2x2.5 MHz bandwidth or 2300 MHz with 2x15 MHz bandwidth, both using FDD. For 5G, 700 MHz is the main low band frequency using 2x5 MHz bandwidth for paired channels in FDD (MIMO 4x4). In contrast, 5G can also take advantage of Time Division Duplexing (TDD) spectrum at 3.5 GHz (MIMO 4x4) with a single 1x50 MHz bandwidth channel, with a 4:1 ratio between the downlink to uplink, given the download capacity is the bottleneck in cellular systems. In terms of other parameters, the MNO administration cost is treated as 20% of the network and the corporation tax rate is treated as 22% of profit, as is the baseline rate in India. The prevailing Weighted Average Cost of Capital (WACC) for India is 10% [161]. Having detailed how the generalizable model is adapted for India's case study example, the results will now be reported. VI. Results
The viability of 4G and 5G technologies in delivering universal broadband over the study period are visualized in Figure 5 in the different scenarios and strategies. The cumulative cost is used to demonstrate the required investment needed to provide coverage up to each population decile (with deciles sorted based on population density). Across the scenarios tested, the results demonstrate that the capacity per user is well correlated with the cost of provision, given the required investment increases significantly as the scenarios become more ambitious. Indeed, as the number of required cell sites increases to serve higher demand, this has a major impact on the cost of building fiber connections, with both 4G and 5G fiber-based strategies being the most expensive options. When interpreting the performance of the different strategies tested, the cumulative cost should be compared relative to the cumulative revenue as this demonstrates the level of viability present. In Scenario 1, we can see that both 4G and 5G using wireless backhauls are viable to service 100% of the population, thus delivering universal broadband. In contrast, fiber strategies can only viably serve up to 50% of the population. In Scenario 2, 4G and a wireless backhaul can viably provide universal coverage of 100% of the population, performing much better than the other options, including the next most competitive technology using a wireless backhaul, 5G NSA. This is due to the existing advantage that 4G has in the baseline availability, in that there are already a substantial number of sites with this technology in use. In contrast, while 5G is more spectrally efficient, all sites need to be upgraded with this new technology. Finally, in Scenario 3 when trying to deliver up to 100 Mbps per user, most strategies are unviable as this target is too ambitious given the potential APRU. However, the cost composition of the required investment is different depending on the deployment context, as demonstrated in Figure 6 for each scenario and strategy. There are two main differences visible. Firstly, the proportion that the backhaul cost contributes to the overall investment composition is high in both the most populated deciles and the least populated deciles. In the former, this is the result of needing lots of sites. Whereas in the latter, this is the result of the backhaul needing to traverse a longer distance to the closest fiber PoP. Secondly, the proportion that the spectrum cost contributes varies. In more populated areas, there is a much higher contribution to the cost of the overall spectrum (because of the greater population), whereas, in the final less populated deciles (where there are fewer people), the contribution to the overall spectrum cost is much lower. These two factors lead to an observable pattern across the scenarios and strategies tested. The aggregate cost per decile is generally higher in both the most and least populated areas, whereas the aggregate cost is lower in the middle deciles. Aggregate costs overlook the number of users served per decile, therefore in Figure 7, the required investment is broken down per user. Again, the results are reported by the cost type for each decile across the different scenarios and strategies. There is a strong relationship across the distribution, whereby the cost per user is lower in the first population deciles, where the population density is highest. The cost per user then increases in tandem with the decrease in population density. In Figure 7, it is also useful to view the required cost per user by decile for the study period because this is a much more meaningful number, given monthly and annual ARPU is generally well understood because many people have cellular subscriptions.
VOLUME XX, 2021 Figure 6 Required investment by population decile for each scenario, strategy, and cost type
VOLUME XX, 2021 Figure 7 Per user cost by population decile for each scenario, strategy, and cost type
VOLUME XX, 2021 Figure 8 The impact of spectrum costs Even with 4G with a wireless backhaul, we can see that $451-850 per user in the most rural decile is going to be challenging, and thankfully the comparative cost for 5G NSA with a wireless backhaul is lower at $317-483 across the scenarios. Both RAN technologies using fiber are far too expensive for the hardest-to-reach areas, with the cost ranging from $1583-3027 for 4G and $1072-1625 for 5G NSA. With spectrum playing a large part in the cost composition of the cheapest technology options, such as 4G and 5G NSA both using a wireless backhaul, it is worth investigating the impact of changes in spectrum prices on the viability of deployment. In Figure 8, the cumulative revenue across population deciles is plotted against the baseline, as well as different decreases in spectrum prices. The aim is to evaluate the impact of spectrum price reductions as they filter through into the cumulative cost of deployment against the point at which the cost curve crosses the cumulative revenue. If a particular decile has a revenue curve above the cost curve the scenario and strategy are viable. In contrast, if
VOLUME XX, 2021 the cost is above the revenue, then the scenario and strategy are unviable. Viability naturally varies across the different scenarios and strategies. With lower capacity per user, such as in Scenario 1, most strategies are either fully viable or close to fully viable with the baseline spectrum price, except for 4G with a fiber backhaul. However, delivering a minimum speed of 25 Mbps in urban areas and 2 Mbps in rural areas, may be perceived as not ambitious enough. Thus, in Scenario 2, the available capacity is an improvement, but viability already becomes difficult without resulting to using either wireless backhaul or reducing the spectrum price. For example, 5G NSA with a fiber backhaul is unviable in the baseline, but if spectrum prices were eliminated altogether it would be possible to viably reach 100% population coverage (although, this may not be politically a feasible option and would only be plausible if a universal service obligation was introduced to guarantee delivery). With the most ambitious target in Scenario 3, all strategies are unviable in the baseline. Even with a drastic reduction in spectrum prices, fiber backhaul options are still unviable in all circumstances. There are important results to take note of in Scenario 3, however. For example, a 60% reduction in spectrum price would enable 4G with a wireless backhaul to become viable in reaching full population coverage, or more importantly, a 40% reduction in the case of 5G NSA with a wireless backhaul would enable coverage to reach 100% of the population. VII. DISCUSSION
The assessment presented in the analytical part of this paper used an open-source modeling codebase to quantitatively evaluate a range of 4G and 5G universal broadband strategies. A combination of remote sensing and infrastructure simulation techniques were combined to provide insight into the capacity, coverage, and cost of both 4G and 5G infrastructure strategies. The results provide insight into the viability of different strategies, depending on heterogenous per user capacity scenarios, including providing the required investment on a per user basis. Finally, a sensitivity analysis was performed to quantify the impact that governmental spectrum pricing regimes have on the economics of universal broadband connectivity, with ramifications for both short term deployment and long-term evolution to 6G. This section now discusses their ramifications regarding the first two research questions articulated in the introductory section of this paper. The first question for investigation was as follows:
How do different 4G and 5G strategies perform in delivering universal broadband coverage?
In terms of the performance of the strategies across the scenarios, the required investment for universal broadband increased as the ambition of the user capacity scenario grew. Generally, the fiber backhaul strategies were much more expensive, supporting the idea that wireless backhaul will remain a key approach for delivering 4G and 5G universal broadband over the coming decade for hard-to-serve areas, should there be no changes in the fiscal landscape. For example, 100% of the population could viably be served in Scenario 1 (2-25 Mbps) using both 4G and 5G with wireless backhaul, whereas fiber strategies could only serve up to 50% viably. Moreover, total population coverage could be achieved in Scenario 2 (5-50 Mbps) using 4G with a wireless backhaul. However, in all circumstances Scenario 3 (10-100 Mbps) was unviable regardless of the strategy as this target is too ambitious given the potential APRU, which can be very low for rural areas. The aggregate cost across the deciles modeled exhibited a U-shape pattern. Hence, there was a much higher aggregate cost in both the most and least populated areas but a considerably lower aggregate cost in the middle deciles where the population density is much more amenable to deploying low-cost 4G and 5G broadband connectivity. When considering the required investment per user, there was a strong dynamic where the cost per user was lower in the deciles with the highest population densities, but as the population density decreased, the cost per user inversely increased. This results from scale economies and the need to split the fixed costs in cellular deployment over the local users accessing specific infrastructure connections. This is not unique to cellular and is exhibited in all infrastructure networks, such as transportation, energy, and water. To provide universal broadband connectivity, we know the most considerable challenge will be in serving the hardest-to-reach areas with the lowest population density. The results show that the costs differ in serving the final population decile depending on the technology deployed. For example, with 4G using a wireless backhaul, the cost per user in the most rural decile was between $451-850 across the different scenarios. Given how low incomes can be in rural areas, this is by no means an easy target to reach using market methods alone, and state subsidies may be required to provide a subsidy for unviable areas. Fortunately, deploying 5G NSA with a wireless backhaul is the cheapest option in these situations, with the cost per user ranging between $317-483 across the scenarios. This compared with much larger per user costs using fiber, where the investment
VOLUME XX, 2021 would need to range from $1583-3027 for 4G and $1072-1625 for 5G NSA across the scenarios tested. However, the caveat to any 5G strategy would be whether the local population had 5G-enabled handsets to take advantage of the available capacity. Having discussed the first research question, the second will now be evaluated, which was as follows: How do spectrum prices affect the performance of different coverage-focused universal broadband strategies?
Governments have many instruments at their disposal to help reduce the costs of broadband deployment in the hope of achieving universal coverage. High spectrum prices are a well-known issue, particularly for India, the example country assessed here. Therefore, the use of sensitivity analysis for this model parameter helps provide insight into the ramifications of potential policy changes. As the least ambitious scenario (2-25 Mbps) was either viable or close to viable for most 4G and 5G strategies, there is less relevance here in exploring spectrum price changes, especially as policy ambitions might be aiming higher than the user capacities targeted in this option. However, in Scenario 2 (5-50 Mbps), while 4G and 5G using wireless backhaul was viable for providing universal broadband, there were other interesting results. 4G with fiber was not viable, even with reduced spectrum costs, but 5G NSA with fiber could be plausibly delivered universally if the spectrum cost was eliminated . This would obviously take significant political will to make such a bold move and would require affiliated coverage obligations to ensure MNOs deliver the necessary infrastructure but could provide a significant improvement for the availability of broadband connectivity, and also provide a fantastic starting point for evolving to 6G, where fiber backhaul is almost certainly going to be required. Finally, Scenario 3 (10-100 Mbps) provides much more admirable per user capacity. Therefore, it is attractive that only a 40% spectrum price reduction would viably enable 5G NSA using wireless backhaul to provide universal broadband to 100% of the population, under the engineering and economic conditions assessed here. Having discussed the ramifications of the results for the 4G and 5G universal broadband assessment undertaken, the conclusion will now consider the broader implications, particularly with reference over the long term to universal 6G.
VIII. CONCLUSION
Can conclusions be developed to inform current 5G policy and future 6G standardization and deployment? For example, what do these results mean for universal broadband policy? Are there implications for the 6G R&D effort? Indeed, which issues should engineers researching 6G technologies be cognizant of to achieve our shared aim of universal broadband connectivity? These important questions will now be answered by drawing relevant conclusions, helping to answer the third research question articulated in the introduction of the paper.
The technology choices currently being made have major long-term trade-offs . While this may sound platitudinous, this analysis demonstrates that MNOs and governments need to be aware of how backhaul decisions will play out over the next decade and beyond. For example, wireless backhaul methods are clearly the winner in helping to achieve expedited cost-efficient deployment of broadband connectivity in hard-to-reach rural and remote areas. However, if we work from the assumption that fiber is almost certainly going to be required to deploy high quality broadband connectivity, for example via universal 6G, governments need to be aware that it may make more economic sense to deploy fiber now, rather than wireless. Obviously, this takes resources but as the analysis in this assessment has shown, the spectrum revenues extracted from the telecom sector are significant and changes to this framework would enable greater fiber penetration to help deliver broadband connectivity. For example, universal 5G using fiber backhaul could be achieved merely by eliminating the spectrum cost. While this is a politically sensitive question (as spectrum revenues are alluring for governments), the real issue is the potential benefits gained from providing enhanced broadband connectivity. Indeed, if they outweigh the revenues generated via spectrum licensing then they may warrant a re-evaluation of the current strategy by government. This issue begins to touch on the next conclusion.
Current broadband strategies based on 4G or 5G generally overlook temporal evolution . This is to the detriment of achieving long-term objectives. Government regulators of telecom markets usually take a relatively short-term perspective (~3-years) on their decisions for the various broadband strategies employed. Our conjecture, informed by the findings of this analysis, is that this type of short-term horizon is too limited. Thus, there needs to be greater appreciation for how cellular infrastructure will be upgraded as each generation is deployed, for example, from 4G to 5G to 6G. This is not to say governments should attempt to predict or forecast the market or indeed technological development for telecom technologies. Instead, there should be greater recognition that telecom regulators can introduce infrastructure policies which encourage the deployment of
VOLUME XX, 2021 favorable technologies which will provide long-term benefits. In the case of the assessment presented in this paper, an example would be developing supportive policies which encourage greater fiber deployment. Fiber in the ground, that can be easily accessed by MNOs and other telecom providers, will have long-term benefits. Indeed, those benefits are well documented, with society developing considerably when citizens having greater opportunities to use digital technologies. Moreover, the economy benefits from efficient infrastructure in terms of greater productivity improvements and how this contributes to growth in a nation’s Gross Domestic Product (which in turn generates greater tax revenue). Universal broadband is fundamentally a good thing, but we need to consider the evolution over time between generations of technology.
6G R&D efforts need to remember the other cost factors that will influence global broadband coverage . In 5G many new and fantastic ways to deliver higher capacity were introduced, and in turn help to reduce the cost per bit of data transfer (e.g. 64x64 Massive MIMO). However, this is one example of a uniquely dense urban solution for providing capacity. In fact, 5G in general did very little to help deploy broadband for rural and hard-to-reach areas. Granted, some research groups did undertake efforts on this topic but generally it was a small-scale activity, focusing mainly on rural deployment. Thankfully, many have already recognized the limitations of 5G in this regard and have attempted to bring this up the agenda for 6G R&D and future standardization. This is no doubt highly important, and the assessment carried out in this paper supports that approach, while also wishing to contribute conclusions of our own. The challenge will be in helping to deploy wide-area connectivity solutions in low-APRU environments which are able to maximize efficiency in terms of spectrum and energy use, and therefore cost.
There needs to be a greater emphasis on independent open-source cost assessment for candidate 6G technologies in earlier phases of standardization . In many ways the cost assessment of 5G technologies was very much an afterthought. This mistake must not be repeated, and without undertaking independent assessment of these technologies in advance, 6G will fall into the same position. Many of the standardized technologies were a set of very urban solutions, rather than the engineering community presenting technological options for a wide range of urban and rural connectivity problems. Moreover, the 5G standardization process lacked the use of open-source tools widely used across the software development community, but which would hugely help identify the best technological candidates for standardization. More work should be openly published which evaluates the use of different network architectures in heterogenous deployment scenarios. This should provide compelling evidence for researchers to help support those technologies which provide the best solutions in terms of cost-efficiency.
Having identified four key conclusions, future research will now be discussed. Firstly, there needs to be more assessment evaluating the trade-off in cost for remote areas between 5G cellular and newly deployed Low Earth Orbit (LEO) satellite broadband constellations, such as those being launched by Space X (Starlink), OneWeb and Blue Origin (Kuiper). Given the latency provided by LEO broadband satellites is now highly competitive with terrestrial options, it may be more affordable to use this connectivity to provide small, single villages with connections where the number of residents is under the viable threshold for cellular technologies to be deployed. Secondly, there also needs to be more assessment evaluating the size of the benefits from enhanced digital connectivity because this would help more robust cost-benefit assessment in government be undertaken in relation to the provision of reliable broadband connectivity.
ACKNOWLEDGMENT
The authors would like to thank their respective institutions for funding support, as well as anonymous reviewers of the paper.
REFERENCES [1] D. Kalbande, Z. khan, S. Haji, and R. Haji, ‘6G-Next Gen Mobile Wireless Communication Approach’, in 2019 3rd International conference on Electronics, Communication and Aerospace Technology (ICECA), Jun. 2019, pp. 1–6, doi: 10.1109/ICECA.2019.8821934. [2] C. Li, W. Guo, S. C. Sun, S. Al-Rubaye, and A. Tsourdos, ‘Trustworthy Deep Learning in 6G-Enabled Mass Autonomy: From Concept to Quality-of-Trust Key Performance Indicators’, IEEE Vehicular Technology Magazine, vol. 15, no. 4, pp. 112–121, Dec. 2020, doi: 10.1109/MVT.2020.3017181. [3] W. Guo, ‘Explainable Artificial Intelligence for 6G: Improving Trust between Human and Machine’, IEEE Communications Magazine, vol. 58, no. 6, pp. 39–45, Jun. 2020, doi: 10.1109/MCOM.001.2000050. [4] A. Bajpai and A. Balodi, ‘Role of 6G Networks: Use Cases and Research Directions’, in 2020 IEEE Bangalore Humanitarian Technology Conference (B-HTC), Oct. 2020, pp. 1–5, doi: 10.1109/B-HTC50970.2020.9298017. [5] C. L. Stergiou, K. E. Psannis, and B. B. Gupta, ‘IoT-based Big Data secure management in the Fog over a
VOLUME XX, 2021
6G Wireless Network’, IEEE Internet of Things Journal, pp. 1–1, 2020, doi: 10.1109/JIOT.2020.3033131. [6] S. Aggarwal, N. Kumar, and S. Tanwar, ‘Blockchain Envisioned UAV Communication Using 6G Networks: Open issues, Use Cases, and Future Directions’, IEEE Internet of Things Journal, pp. 1–1, 2020, doi: 10.1109/JIOT.2020.3020819. [7] S. P. Rout, ‘6G Wireless Communication: Its Vision, Viability, Application, Requirement, Technologies, Encounters and Research’, in 2020 11th International Conference on Computing, Communication and Networking Technologies (ICCCNT), Jul. 2020, pp. 1–8, doi: 10.1109/ICCCNT49239.2020.9225680. [8] T. Vrind, S. Rao, L. Pathak, and D. Das, ‘Deep Learning-based LAP Deployment and Aerial Infrastructure Sharing in 6G’, in 2020 IEEE International Conference on Electronics, Computing and Communication Technologies (CONECCT), Jul. 2020, pp. 1–5, doi: 10.1109/CONECCT50063.2020.9198319. [9] R. Gupta, D. Reebadiya, and S. Tanwar, ‘6G-enabled Edge Intelligence for Ultra -Reliable Low Latency Applications: Vision and Mission’, Computer Standards & Interfaces, p. 103521, Feb. 2021, doi: 10.1016/j.csi.2021.103521. [10] R. Gupta, A. Nair, S. Tanwar, and N. Kumar, ‘Blockchain-assisted secure UAV communication in 6G environment: Architecture, opportunities, and challenges’, IET Communications, vol. n/a, no. n/a, 2021, doi: https://doi.org/10.1049/cmu2.12113. [11] S. Goyal, N. Sharma, I. Kaushik, B. Bhushan, and N. Kumar, ‘A Green 6G Network Era: Architecture and Propitious Technologies’, in Data Analytics and Management, Singapore, 2021, pp. 59–75, doi: 10.1007/978-981-15-8335-3_7. [12] C. Huang et al., ‘Holographic MIMO Surfaces for 6G Wireless Networks: Opportunities, Challenges, and Trends’, IEEE Wireless Communications, vol. 27, no. 5, pp. 118–125, Oct. 2020, doi: 10.1109/MWC.001.1900534. [13] Z. Lv and N. Kumar, ‘Software defined solutions for sensors in 6G/IoE’, Computer Communications, vol. 153, pp. 42–47, Mar. 2020, doi: 10.1016/j.comcom.2020.01.060. [14] H. Yang, A. Alphones, Z. Xiong, D. Niyato, J. Zhao, and K. Wu, ‘Artificial-Intelligence-Enabled Intelligent 6G Networks’, IEEE Network, vol. 34, no. 6, pp. 272–280, Nov. 2020, doi: 10.1109/MNET.011.2000195. [15] J. C.-W. Lin, G. Srivastava, Y. Zhang, Y. Djenouri, and M. Aloqaily, ‘Privacy Preserving Multi-Objective Sanitization Model in 6G IoT Environments’, IEEE Internet of Things Journal, pp. 1–1, 2020, doi: 10.1109/JIOT.2020.3032896. [16] E. Yaacoub and M. Alouini, ‘A Key 6G Challenge and Opportunity—Connecting the Base of the Pyramid: A Survey on Rural Connectivity’, Proceedings of the IEEE, vol. 108, no. 4, pp. 533–582, Apr. 2020, doi: 10.1109/JPROC.2020.2976703. [17] S. Dang, O. Amin, B. Shihada, and M.-S. Alouini, ‘What should 6G be?’, Nat Electron, vol. 3, no. 1, pp. 20–29, Jan. 2020, doi: 10.1038/s41928-019-0355-6. [18] H. Saarnisaari et al., A 6G White Paper on Connectivity for Remote Areas. 2020. [19] A. Chaoub et al., 6G for Bridging the Digital Divide: Wireless Connectivity to Remote Areas. 2020. [20] J. Bhat and S. AlQahtani, ‘6G Ecosystem: Current Status and Future Perspective’, IEEE Access, vol. PP, pp. 1–1, Jan. 2021, doi: 10.1109/ACCESS.2021.3054833. [21] L. Bariah et al., ‘A Prospective Look: Key Enabling Technologies, Applications and Open Research Topics in 6G Networks’, IEEE Access, vol. 8, pp. 174792–174820, 2020, doi: 10.1109/ACCESS.2020.3019590. [22] Y. Lu and X. Zheng, ‘6G: A survey on technologies, scenarios, challenges, and the related issues’, Journal of Industrial Information Integration, vol. 19, p. 100158, Sep. 2020, doi: 10.1016/j.jii.2020.100158. [23] X. You et al., ‘Towards 6G wireless communication networks: vision, enabling technologies, and new paradigm shifts’, Sci. China Inf. Sci., vol. 64, no. 1, p. 110301, Nov. 2020, doi: 10.1007/s11432-020-2955-6. [24] R. S. Fish, ‘Universal Broadband Service and the Pandemic’, IEEE Communications Standards Magazine, vol. 4, no. 3, pp. 3–3, Sep. 2020, doi: 10.1109/MCOMSTD.2020.9204589. [25] V. Chamola, V. Hassija, V. Gupta, and M. Guizani, ‘A Comprehensive Review of the COVID-19 Pandemic and the Role of IoT, Drones, AI, Blockchain, and 5G in Managing its Impact’, IEEE Access, vol. 8, pp. 90225–90265, 2020, doi: 10.1109/ACCESS.2020.2992341. [26] S. K. A. Kumar, R. Stewart, D. Crawford, and S. Chaudhari, ‘Business model for rural connectivity using multi-tenancy 5G network slicing’, in 2020 IEEE 17th International Conference on Smart Communities: Improving Quality of Life Using ICT, IoT and AI (HONET), Dec. 2020, pp. 182–188, doi: 10.1109/HONET50430.2020.9322837. [27] K. K. Thakur and R. Prasad, ‘Wi-Fi for Affordable Broadband & 5G in Rural Areas’, Journal of Mobile Multimedia, pp. 225-244-225–244, Feb. 2021, doi: 10.13052/jmm1550-4646.171312. [28] G. Gui, M. Liu, F. Tang, N. Kato, and F. Adachi, ‘6G: Opening New Horizons for Integration of Comfort, Security, and Intelligence’, IEEE Wireless Communications, vol. 27, no. 5, pp. 126–132, Oct. 2020, doi: 10.1109/MWC.001.1900516. [29] M. Matinmikko-Blue et al., ‘White Paper on 6G Drivers and the UN SDGs’, arXiv:2004.14695 [cs,
VOLUME XX, 2021 eess], Apr. 2020, Accessed: Feb. 02, 2021. [Online]. Available: http://arxiv.org/abs/2004.14695. [30] M. Matinmikko-Blue, S. Yrjölä, and P. Ahokangas, ‘Spectrum Management in the 6G Era: The Role of Regulation and Spectrum Sharing’, in 2020 2nd 6G Wireless Summit (6G SUMMIT), Mar. 2020, pp. 1–5, doi: 10.1109/6GSUMMIT49458.2020.9083851. [31] G. Gür, ‘Expansive networks: Exploiting spectrum sharing for capacity boost and 6G vision’, Journal of Communications and Networks, vol. 22, no. 6, pp. 444–454, Dec. 2020, doi: 10.23919/JCN.2020.000037. [32] A. Dogra, R. K. Jha, and S. Jain, ‘A Survey on beyond 5G network with the advent of 6G: Architecture and Emerging Technologies’, IEEE Access, pp. 1–1, 2020, doi: 10.1109/ACCESS.2020.3031234. [33] R. K. Saha, ‘Licensed Countrywide Full-Spectrum Allocation: A New Paradigm for Millimeter-Wave Mobile Systems in 5G/6G Era’, IEEE Access, vol. 8, pp. 166612–166629, 2020, doi: 10.1109/ACCESS.2020.3023342. [34] I. F. Akyildiz, A. Kak, and S. Nie, ‘6G and Beyond: The Future of Wireless Communications Systems’, IEEE Access, vol. 8, pp. 133995–134030, 2020, doi: 10.1109/ACCESS.2020.3010896. [35] M. Polese, J. M. Jornet, T. Melodia, and M. Zorzi, ‘Toward End-to-End, Full-Stack 6G Terahertz Networks’, IEEE Communications Magazine, vol. 58, no. 11, pp. 48–54, Nov. 2020, doi: 10.1109/MCOM.001.2000224. [36] N. Rajatheva et al., ‘White Paper on Broadband Connectivity in 6G’, arXiv:2004.14247 [eess], Apr. 2020, Accessed: Feb. 13, 2021. [Online]. Available: http://arxiv.org/abs/2004.14247. [37] Z. Frias, L. Mendo, and E. J. Oughton, ‘How Does Spectrum Affect Mobile Network Deployments? Empirical Analysis Using Crowdsourced Big Data’, IEEE Access, vol. 8, pp. 190812–190821, 2020, doi: 10.1109/ACCESS.2020.3031963. [38] H. Viswanathan and P. E. Mogensen, ‘Communications in the 6G Era’, IEEE Access, vol. 8, pp. 57063–57074, 2020, doi: 10.1109/ACCESS.2020.2981745. [39] G. Wikström et al., ‘Challenges and Technologies for 6G’, in 2020 2nd 6G Wireless Summit (6G SUMMIT), Mar. 2020, pp. 1–5, doi: 10.1109/6GSUMMIT49458.2020.9083880. [40] V. Ziegler and S. Yrjola, ‘6G Indicators of Value and Performance’, in 2020 2nd 6G Wireless Summit (6G SUMMIT), Mar. 2020, pp. 1–5, doi: 10.1109/6GSUMMIT49458.2020.9083885. [41] E. J. Oughton and T. Russell, ‘The importance of spatio-temporal infrastructure assessment: Evidence for 5G from the Oxford–Cambridge Arc’, Computers, Environment and Urban Systems, vol. 83, p. 101515, Sep. 2020, doi: 10.1016/j.compenvurbsys.2020.101515. [42] C. Sergiou, M. Lestas, P. Antoniou, C. Liaskos, and A. Pitsillides, ‘Complex Systems: A Communication Networks Perspective Towards 6G’, IEEE Access, vol. 8, pp. 89007–89030, 2020, doi: 10.1109/ACCESS.2020.2993527. [43] V. Ziegler, H. Viswanathan, H. Flinck, M. Hoffmann, V. Räisänen, and K. Hätönen, ‘6G Architecture to Connect the Worlds’, IEEE Access, vol. 8, pp. 173508–173520, 2020, doi: 10.1109/ACCESS.2020.3025032. [44] M. Z. Chowdhury, M. Shahjalal, S. Ahmed, and Y. M. Jang, ‘6G Wireless Communication Systems: Applications, Requirements, Technologies, Challenges, and Research Directions’, IEEE Open Journal of the Communications Society, vol. 1, pp. 957–975, 2020, doi: 10.1109/OJCOMS.2020.3010270. [45] M. Giordani, M. Polese, M. Mezzavilla, S. Rangan, and M. Zorzi, ‘Toward 6G Networks: Use Cases and Technologies’, IEEE Communications Magazine, vol. 58, no. 3, pp. 55–61, Mar. 2020, doi: 10.1109/MCOM.001.1900411. [46] F. Tariq, M. R. A. Khandaker, K.-K. Wong, M. A. Imran, M. Bennis, and M. Debbah, ‘A Speculative Study on 6G’, IEEE Wireless Communications, vol. 27, no. 4, pp. 118–125, Aug. 2020, doi: 10.1109/MWC.001.1900488. [47] L. U. Khan, I. Yaqoob, M. Imran, Z. Han, and C. S. Hong, ‘6G Wireless Systems: A Vision, Architectural Elements, and Future Directions’, IEEE Access, vol. 8, pp. 147029–147044, 2020, doi: 10.1109/ACCESS.2020.3015289. [48] S. J. Nawaz, S. K. Sharma, S. Wyne, M. N. Patwary, and M. Asaduzzaman, ‘Quantum Machine Learning for 6G Communication Networks: State-of-the-Art and Vision for the Future’, IEEE Access, vol. 7, pp. 46317–46350, 2019, doi: 10.1109/ACCESS.2019.2909490. [49] Z. Zhang et al., ‘6G Wireless Networks: Vision, Requirements, Architecture, and Key Technologies’, IEEE Vehicular Technology Magazine, vol. 14, no. 3, pp. 28–41, Sep. 2019, doi: 10.1109/MVT.2019.2921208. [50] Y. L. Lee, D. Qin, L.-C. Wang, and G. H. Sim, ‘6G Massive Radio Access Networks: Key Applications, Requirements and Challenges’, IEEE Open Journal of Vehicular Technology, vol. 2, pp. 54–66, 2021, doi: 10.1109/OJVT.2020.3044569. [51] P. K. Padhi and F. Charrua‐Santos, ‘6G Enabled Industrial Internet of Everything: Towards a Theoretical Framework’, Applied System Innovation, vol. 4, p. 11, Feb. 2021, doi: 10.3390/asi4010011. [52] S. Chen, Y. Liang, S. Sun, S. Kang, W. Cheng, and M. Peng, ‘Vision, Requirements, and Technology Trend of 6G: How to Tackle the Challenges of VOLUME XX, 2021 System Coverage, Capacity, User Data-Rate and Movement Speed’, IEEE Wireless Communications, vol. 27, no. 2, pp. 218–228, Apr. 2020, doi: 10.1109/MWC.001.1900333. [53] I. Tomkos, D. Klonidis, E. Pikasis, and S. Theodoridis, ‘Toward the 6G Network Era: Opportunities and Challenges’, IT Professional, vol. 22, no. 1, pp. 34–38, Jan. 2020, doi: 10.1109/MITP.2019.2963491. [54] M. H. Alsharif, A. H. Kelechi, M. A. Albreem, S. A. Chaudhry, M. S. Zia, and S. Kim, ‘Sixth Generation (6G) Wireless Networks: Vision, Research Activities, Challenges and Potential Solutions’, Symmetry, vol. 12, no. 4, Art. no. 4, Apr. 2020, doi: 10.3390/sym12040676. [55] Y. Liu, X. Yuan, Z. Xiong, J. Kang, X. Wang, and D. Niyato, ‘Federated learning for 6G communications: Challenges, methods, and future directions’, China Communications, vol. 17, no. 9, pp. 105–118, Sep. 2020, doi: 10.23919/JCC.2020.09.009. [56] R. Frieden, ‘The evolving 5G case study in spectrum management and industrial policy’, Telecommunications Policy, vol. 43, no. 6, pp. 549–562, Jul. 2019, doi: 10.1016/j.telpol.2019.04.001. [57] M. Kim, H. Lee, and J. Kwak, ‘The changing patterns of China’s international standardization in ICT under techno-nationalism: A reflection through 5G standardization’, International Journal of Information Management, vol. 54, p. 102145, Oct. 2020, doi: 10.1016/j.ijinfomgt.2020.102145. [58] W. Lemstra, ‘Leadership with 5G in Europe: Two contrasting images of the future, with policy and regulatory implications’, Telecommunications Policy, vol. 42, pp. 587–611, 2018, doi: 10.1016/j.telpol.2018.02.003. [59] E. J. Oughton and Z. Frias, ‘The cost, coverage and rollout implications of 5G infrastructure in Britain’, Telecommunications Policy, vol. 42, no. 8, pp. 636–652, Sep. 2018, doi: 10.1016/j.telpol.2017.07.009. [60] L. Wang, F. Jia, L. Chen, Q. Xu, and X. Lin, ‘Exploring the dependence structure among Chinese firms in the 5G industry’, Industrial Management & Data Systems, vol. ahead-of-print, no. ahead-of-print, Jan. 2021, doi: 10.1108/IMDS-06-2020-0323. [61] J. Xia, ‘China’s telecommunications evolution, institutions, and policy issues on the eve of 5G: A two-decade retrospect and prospect’, Telecommunications Policy, vol. 41, no. 10, pp. 931–947, Nov. 2017, doi: 10.1016/j.telpol.2016.11.003. [62] C. Jeon, S. H. Han, H. J. Kim, and S. Kim, ‘The effect of government 5G policies on telecommunication operators’ firm value: Evidence from China’, Telecommunications Policy, vol. In-Press, 2020, doi: 10.1016/j.telpol.2020.102040. [63] C. H. Kwan, ‘The China–US Trade War: Deep-Rooted Causes, Shifting Focus and Uncertain Prospects’, Asian Economic Policy Review, vol. 15, no. 1, pp. 55–72, 2020, doi: https://doi.org/10.1111/aepr.12284. [64] M. Cave, ‘How disruptive is 5G?’, Telecommunications Policy, vol. 42, no. 8, pp. 653–658, Sep. 2018, doi: 10.1016/j.telpol.2018.05.005. [65] M. Massaro, ‘Next generation of radio spectrum management: Licensed shared access for 5G’, 2017, doi: 10.1016/j.telpol.2017.04.003. [66] M. Matinmikko, M. Latva-aho, P. Ahokangas, and V. Seppänen, ‘On regulations for 5G: Micro licensing for locally operated networks’, Telecommunications Policy, vol. 42, no. 8, pp. 622–635, Sep. 2018, doi: 10.1016/j.telpol.2017.09.004. [67] R. Borralho, A. Mohamed, A. Quddus, P. Vieira, and R. Tafazolli, ‘A Survey on Coverage Enhancement in Cellular Networks: Challenges and Solutions for Future Deployments’, IEEE Communications Surveys Tutorials, pp. 1–1, 2021, doi: 10.1109/COMST.2021.3053464. [68] I. A. Gedel and N. I. Nwulu, ‘Low Latency 5G Distributed Wireless Network Architecture: A Techno-Economic Comparison’, Inventions, vol. 6, no. 1, Art. no. 1, Mar. 2021, doi: 10.3390/inventions6010011. [69] T. Hoeschele, C. Dietzel, D. Kopp, F. H. P. Fitzek, and M. Reisslein, ‘Importance of Internet Exchange Point (IXP) infrastructure for 5G: Estimating the impact of 5G use cases’, Telecommunications Policy, vol. 45, no. 3, p. 102091, Apr. 2021, doi: 10.1016/j.telpol.2020.102091. [70] W. Lehr, F. Queder, and J. Haucap, ‘5G: A new future for Mobile Network Operators, or not?’, Telecommunications Policy, vol. 45, no. 3, p. 102086, Apr. 2021, doi: 10.1016/j.telpol.2020.102086. [71] S. O. Oladejo and O. E. Falowo, ‘Latency-Aware Dynamic Resource Allocation Scheme for Multi-Tier 5G Network: A Network Slicing-Multitenancy Scenario’, IEEE Access, vol. 8, pp. 74834–74852, 2020, doi: 10.1109/ACCESS.2020.2988710. [72] F. Spinelli and V. Mancuso, ‘Towards enabled industrial verticals in 5G: a survey on MEC-based approaches to provisioning and flexibility’, IEEE Communications Surveys Tutorials, pp. 1–1, 2020, doi: 10.1109/COMST.2020.3037674. [73] S. Yrjola, ‘Technology Antecedents of the Platform-Based Ecosystemic Business Models beyond 5G’, in 2020 IEEE Wireless Communications and Networking Conference Workshops (WCNCW), Apr. 2020, pp. 1–8, doi: 10.1109/WCNCW48565.2020.9124823. [74] W. Lambrechts and S. Sinha, ‘The Role of Millimeter-Wave and 5G in the Fourth Industrial Revolution’, in Millimeter-wave Integrated Technologies in the Era of the Fourth Industrial Revolution, W. Lambrechts and S. Sinha, Eds. Cham: Springer International Publishing, 2021, pp. 1–48.
VOLUME XX, 2021 VOLUME XX, 2021 VOLUME XX, 2021 VOLUME XX, 2021 EDWARD J. OUGHTON received the M.Phil. and Ph.D. degrees from Clare College, at the University of Cambridge, U.K., in 2010 and 2015, respectively. He later held research positions at both Cambridge and Oxford. He is currently an Assistant Professor in the College of Science at George Mason University, Fairfax, VA, USA, developing open-source research software to analyze digital infrastructure deployment strategies. He received the Pacific Telecommunication Council Young Scholars Award in 2019, Best Paper Award 2019 from the Society of Risk Analysis, and the TPRC48 Charles Benton Early Career Scholar Award 2021.
ASHUTOSH JHA completed his Ph.D. from the Indian Institute of Management, Calcutta in 2020, and his B.E. (Hons.) in Electrical and Electronics Engineering from BITS-Pilani, Goa, India, in 2008. He is currently an Assistant Professor in the Information Management Group at SPJIMR, Mumbai. His research interests include Adoption and Diffusion of Emerging Technologies, Economics of Next Generation Networks, and Development Informatics. He received the Best Paper award in the 13th International Conference on Digital Society and eGovernments (ICDS 2019), Athens, Greece.completed his Ph.D. from the Indian Institute of Management, Calcutta in 2020, and his B.E. (Hons.) in Electrical and Electronics Engineering from BITS-Pilani, Goa, India, in 2008. He is currently an Assistant Professor in the Information Management Group at SPJIMR, Mumbai. His research interests include Adoption and Diffusion of Emerging Technologies, Economics of Next Generation Networks, and Development Informatics. He received the Best Paper award in the 13th International Conference on Digital Society and eGovernments (ICDS 2019), Athens, Greece.