3.2.1 Spectrum Authorization and Usage Options for 5G

Generally, the use of radio frequency spectrum can be authorized in two ways, first by “individual authorization” in the form of awarding licenses, and secondly by “general authorization”, also referred to as license‐exempt or unlicensed. In [6], four different user modes for the operation of 5G radio access systems have been defined, namely the “service dedicated user mode”, the “exclusive user mode”, the “Licensed Shared Access (LSA) user mode”, and the “unlicensed user mode”. The relationship between these user modes and the two authorization schemes is illustrated in the upper part of Figure 3‐1 named “regulatory framework domain”.

Spectrum usage rights awarded by “individual authorization” are exclusive for the license holder at a given location and/or time. The “service dedicated user mode” refers to spectrum designated to services other than public mobile communications, which are indented to be integrated into the 5G ecosystem, for example Intelligent Transport Systems (ITS) or Public Protection and Disaster Relief (PPDR) applications. This spectrum is to be used only for dedicated services and applications. Spectrum designated to public mobile communications falls into the “exclusive user mode”. In the “LSA user mode”, a non‐mobile communications license holder (incumbent) would share spectrum access rights with one (or more) LSA licensee(s), which can use the spectrum under defined conditions subject to an individual agreement and permission by the relevant regulatory authority. These three user modes can occur either in their basic form (see continuous lines in Figure 3‐1) or as an evolution of current approaches in the form of “limited spectrum pool” or “mutual renting” (shown through dashed lines in Figure 3‐1).

In the “limited spectrum pool” usage scenario, a limited number of known operators obtain authorizations to access a spectrum band dynamically. Mutual agreements between these licensees shall guarantee that in the long‐term each participating operator has a predictable minimum value from the shared spectrum. In the “mutual renting” scenario, an operator would rent at least a part of its licensed spectrum resources to another operator, based on mutually agreed rules. Depending on the time period of the spectrum access, the spectrum usage scenarios “limited spectrum pool”, “mutual renting” and “vertical sharing” may be considered as exclusive (static) or shared (dynamic) use.

Spectrum access and usage rights granted by general authorization are covered by the “unlicensed user mode”, also known as license‐exempt usage. This means that such users have no individual license, but the spectrum usage is subject to certain technical restrictions or conditions, for example limited transmission power or mitigation techniques like duty cycle or listen‐before‐talk. In this user mode, spectrum users cannot claim protection from interference by other users.

In the case when spectrum sharing takes place between systems of different priority, this is referred to as “vertical sharing”, whereas spectrum sharing between systems of equal priority is called “horizontal sharing”. For example, wireless access systems (WAS) operating in parts of the 5 GHz spectrum have to avoid interference into incumbent radar systems (vertical sharing), and also have to employ mitigation techniques to coexist with other WAS systems (horizontal sharing).

5G systems are expected to support all spectrum usage scenarios indicated in Figure 3‐1, in order to facilitate high spectrum usage efficiency. In network operations, several scenarios may occur simultaneously.

3.2.2 Requirements for Different 5G Usage Scenarios

5G is going to support diverse use cases and applications, covering not only the traditional services for mobile subscribers, but also applications for a number of vertical industries like the automotive, energy, eHealth or manufacturing sector. All 5G use cases and applications can be assigned to one or more of the following three main usage scenarios introduced in Chapter 2 [7]:

• Enhanced Mobile Broadband (eMBB), addressing human‐centric use cases for access to multi‐media content and data services. This usage scenario embraces a number of use cases and deployment scenarios with quite diverging requirements. For example in a hotspot scenario, extreme high throughputs and low‐latency communications are in the foreground, while for wide area coverage the customer Quality of Experience (QoE) with reliable and moderate data rates over the coverage area is in focus.
• Massive Machine‐Type Communications (mMTC), characterized by wireless connectivity of billions of network‐enabled devices with prioritization on wide area coverage and deep indoor penetration, typically transmitting non‐delay‐sensitive data at low rates. Usage scenarios for mMTC are for example smart cities, smart buildings, or sensor networks for farming and agriculture.
• Ultra‐Reliable and Low Latency Communications (URLLC), having stringent requirements on latency and availability. Examples for URLLC are the wireless automation of production facilities, monitoring of critical infrastructures in a smart grid, remote medical surgery, remote robotics, the tactile Internet, or vehicular traffic efficiency and safety.

eMBB applications require a mixture of frequency bands including lower bands for coverage purposes as well as for low to medium data traffic, and higher bands with large contiguous bandwidths to deal with the expected extremely high traffic demand. Exclusive licensed spectrum is essential to guarantee coverage obligations and a minimum QoS for the customers. Spectrum authorized by other licensing regimes, for example LSA or unlicensed access, is a supplementary option to increase the overall spectrum availability.

mMTC applications mainly demand frequency spectrum below 6 GHz, and spectrum below 1 GHz is needed in particular for wide area coverage and reliable outdoor to indoor penetration. Therefore, exclusive licensed spectrum is the preferred option. However, also higher frequency bands and other licensing regimes might be considered, subject to the specific mMTC application requirements.

URLLC applications require high and reliable spectrum availability. Thus, licensed spectrum is considered most appropriate for these kinds of services. For communications for automotive safety and efficiency, see also Chapter 14, the frequency band 5875‐5925 MHz harmonized for ITS is an option. Particularly for high‐speed vehicles and in rural environments, spectrum below 1 GHz is well suited.

3.3.1 Main Parameters for Spectrum Bandwidth Demand Estimations

The required bandwidth for any specific radiocommunication network, i.e. also the future 5G networks, greatly depends on the following three parameters:

1. The targeted QoS. This parameter may differ to a large extent, depending on the service provided by the network. For 5G, the mix of the three main usage scenarios (eMBB, mMTC and URLLC) needs to be considered. One key aspect to be taken into account for QoS is the prediction of traffic patterns for different services.
2. The area spectral efficiency. This parameter is expressed in bit/s/Hz/cell and describes how efficiently, in terms of the data rate per bandwidth per cell, the available spectrum is used. Innovations currently under standardization or research will greatly impact the achievable area spectral efficiency in 5G and its foreseeable evolution. The area spectral efficiency can be increased by enhancing the achievable single link spectral efficiency within an individual transmission reception point (TRP), for instance through higher order modulation and coding schemes, massive multiple input‐multiple output (MIMO) or interference cancelation technologies, as covered in detail in Chapter 11. Also, the coordination of several TRPs present in the area, for instance in the form of advanced inter‐cell interference coordination (ICIC) and coordinated multi‐point (CoMP) can increase the area spectral efficiency, as detailed in Chapter 12. These latter approaches can strongly benefit from C‐RAN deployments and software‐defined networking (SDN), as discussed in Section 6.8.
3. Physical deployment of network, TRPs and user distributions. The number of network TRPs in high traffic density areas has been substantially increased in current legacy networks, with the provision of different levels of small cells, in order to maintain the QoS with the same amount of spectrum bandwidth. However, this high density deployment has a clear impact on both capital expenditures (CAPEX) and operational expenditures (OPEX) associated with the provision of the service, which could be unsustainable from an economic point of view. Realistic estimations on user density distributions and TRP deployments are needed in order to perform an adequate evaluation of the required spectrum bandwidth. Parameters like the speed of users or moving TRPs will also have an impact on the final achievable spectral efficiency of the associated links and thus the required bandwidth.

In the following sub‐sections, an overview of current approaches and their applicability to 5G is given, and a statistical procedure with a technology and frequency band agnostic approach is introduced.

3.3.2 State of the Art of Spectrum Demand Analysis

Currently, the spectrum requirement analysis for new terrestrial IMT radiocommunication networks such as 5G is evolving from [8]. The parameters for the bandwidth demand analysis introduced in Section 3.3.1 are implemented in this tool in the following manner:

1. The targeted QoS is characterized by the parametrization of the requirements of twenty different service categories. The parameters associated to these service categories are the foreseen traffic models of different services based on market forecasts, defined by: session arrival rate [session/s/user], mean device bit rate [kbps] and mean session duration [s/session].
2. The area spectral efficiency is directly associated with each of the four different TRP layers considered: macro cells, micro cells, pico cells and hot spots, with different values associated to the three considered deployment scenarios (dense urban, sub‐urban and rural). Therefore, each TRP is characterized by the area spectral efficiency [bits/s/Hz/cell] value used in the respective radio access technology (RAT), taking into account the estimated traffic per area.
3. The physical deployment scenarios are characterized by the service environment and by the user density [users/km²]. Depending on the service environment, three main scenarios are considered: dense urban (with three different densities), sub‐urban (with two different densities) and rural. Furthermore, each user in a scenario is associated with a probability of being in one of the following three mobility states: stationary (0‐5 km/h), low mobility (5‐50 km/h), and high mobility (50‐250 km/h).

The results of studies on estimated spectrum requirements for terrestrial IMT (pre‐5G technologies) in the year 2020, as provided in [8], are shown in Table 3‐1.

Spectrum demand estimates for 5G (IMT‐2020) are even higher. For example, with a technical performance‐based framework for spectrum bandwidth demand analysis, a demand of about 7 GHz of additional spectrum in bands above 6 GHz is estimated for a dense urban information society scenario, and more than 14 GHz for a virtual reality office environment [6].

3.3.3 Spectrum Demand Analysis on Localized Scenarios

When planning mobile networks, mobile network operators (MNOs) aim to achieve a targeted QoS using the limited assigned bandwidth within one or several bands, with minimum deployment and operational costs. Usually, advanced radio frequency propagation tools are used, based on ray tracing techniques and thus being capable of evaluating the achievable signal‐to‐interference‐plus‐noise ratio (SINR) once TRP positions, carrier frequency and technology parameters are set.

The SINR values for randomly distributed locations in a TRP coverage area follow a Gaussian curve. Usually, the figure of merit of a new technology is evaluated as the achievable increase of the mean value of this Gaussian curve for any specific scenario. This approach is also used in 3GPP benchmarking. For example, in [9], the performance degradation of the SINR distribution for different channel calibration errors in a 32 TRP scenario (assuming a downlink CoMP joint transmission technique) is described as variations in the Gaussian SINR values.

Based on this principle, a method for the spectrum demand analysis in a limited coverage area, with statistical model assumptions for the SINR has been developed [6]. The main parameters for spectrum bandwidth demand estimations are implemented as follows:

1. The targeted QoS is characterized by the achievable throughput for user sessions, taking into account that for each session type various levels of QoS are established, since services like video streaming may be delivered in different qualities, linked to different compressing levels. Moreover, for each type of user equipment (UE), a different probability to be connected with a different session type (data transfer, video streaming, web browsing, etc.) can be applied by a throughput requirement statistical model.
2. The area spectral efficiency is calculated for each scenario. The TRPs deployed in the scenarios (assuming a heterogenous, layered deployment with macro, micro and pico cells) are characterized ‐ for each frequency band considered in the scenario ‐ by the statistical distribution of the achievable SINR in their coverage area. In order for a UE to be schedulable, the UE‐TRP link must present a spectral efficiency above the value established as the minimum level. The radio links to the various types of UEs considered in the scenarios (at each carrier frequency) are characterized by: a degradation coefficient (from values achievable applying Shannon theorem), the maximum values of spectral efficiency (depending on the number of antennas involved in the MIMO system), and a minimum value for SINR below which the UE is considered out of coverage.
3. The physical deployment scenarios are characterized by the user density (i.e., users/km²) and the TRP positions. Three different levels of user densities are considered: general density over the full scenario, high density spots with increased density (campus, office areas, etc.), and ultra‐dense hot spot areas. The TRPs are deployed according to these three levels, taking into account a realistic deployment of the macro layer, i.e. including random phase and distance errors from the canonical regular grid used in 3GPP evaluations, leading to realistic coverage areas provided by the Voronoi cells (see Figure 3‐2) in accordance with the distance to actual positions of TRPs in the scenario.

Since the achievable QoS may vary greatly depending on the statistical models included in the evaluation tools, a Monte Carlo computational approach is required to get a reliable evaluation of the QoS, achievable with different bandwidths available in different frequency bands.

3.4.1 Bands Identified for IMT and Under Study in ITU‐R

The widespread usage of smartphones and tablets is causing a continuing growth in mobile data communication. According to [10], mobile data traffic will increase with an estimated compound annual growth rate (CAGR) of 47% from 2016 to 2021. In order to cope with such a huge mobile traffic demand, spectrum regulators are working together with involved industries in order to identify bands to make sufficient spectrum available for 5G, focusing on bands that have the potential to be harmonized globally. In consideration of the diverse usage scenarios, technologies and applications enabled by 5G, access to different spectrum bands with different characteristics is required:

• spectrum at lower frequencies to enable coverage of wide areas;
• spectrum at higher frequencies with larger bandwidths to provide necessary capacity and enable higher data rates;
• spectrum at very high frequencies (above 24 GHz) and with very large bandwidths, for providing ultra‐high capacity and data rates.

Concerning spectrum above 24 GHz, the World Radiocommunication Conference in 2015 (WRC‐15) approved by its Resolution 238 [1] to conduct studies in the respective groups within the ITU Radiocommunication sector (ITU‐R), in order to determine spectrum needs and to define sharing and compatibility conditions in the frequency ranges between 24.25 GHz and 86 GHz, as listed in the right column of Table 3‐2. Most of these bands already have an allocation to the mobile service on a primary basis, except the bands 31.8‐33.4 GHz, 40.5‐42.5 GHz and 47‐47.2 GHz, which may require such an allocation in addition.

Based on the results of the studies mentioned above, according to agenda item 1.13, the WRC‐19 will consider the identification of these frequency bands for the future development of International Mobile Telecommunications (IMT), including possible additional allocations to the mobile service on a primary basis.

In order to conduct the appropriate studies, a Task Group (TG 5/1) was established in ITU‐R under the Study Group 5 as being responsible for WRC‐19 agenda item 1.13, while the ITU‐R Working Party 5D was tasked to conduct and complete the studies with regard to spectrum needs, technical and operational characteristics including protection criteria, and deployment scenarios for the terrestrial component of IMT. These studies were completed by March 2017. TG 5/1 is also responsible for the input to the Conference Preparatory Meeting (CPM‐19) concerning WRC‐19 agenda item 1.13.

For the sharing and compatibility studies to be performed, the protection criteria of the radio services which have already a service allocation in the respective band or adjacent to this band have to be taken into account, and all relevant interference scenarios need to be considered. According to the TG 5/1 work plan, the results of the sharing and compatibility studies will be finalized in 2018.

3.4.2 Further Potential Frequency Bands

Beside the frequency bands above 24 GHz under study for IMT‐2020 (see right column of Table 3‐2), spectrum below 6 GHz is required to fulfil the requirements of all potential 5G use cases. For this purpose, bands identified for IMT (see Table 3‐2), but not yet in usage by cellular mobile systems, offer the best opportunities. Examples are the 600/700 MHz bands considered in US and Europe, and the 3.3‐3.8 GHz range from which parts are in focus for 5G in Europe, China, Japan and South Korea. The range 4.4‐4.9 GHz, although predominantly not identified for IMT, is under consideration in Japan and China [11].

Spectrum at 28 GHz is one of the main potential candidates for first deployments of 5G above 24 GHz, as the band 27.5‐28.35 GHz is put into focus for initial 5G commercialization in the US [12] and in other large markets such as Korea (26.5‐29.5 GHz) and Japan (27.5‐29.5 GHz)[11]. There is also some interest from the mobile industry in investigating bands in the range of 6‐24 GHz [4], although corresponding proposals were not supported by WRC‐15.

Applications for which spectrum bands are already harmonized [13] may also be realized with 5G technology, for example traffic safety applications within the band 5875‐5925 MHz in support for the automotive sector, or wireless industrial applications within the band 5725‐5875 MHz for the factories sector. Further applications with already harmonized frequency bands are for instance Public Protection and Disaster Relief (PPDR) or Programme Making and Special Events (PMSE).

3.4.3 5G Roadmaps

The European Commission signed an agreement with the 5G Infrastructure Association, representing major industry players, to establish the 5G PPP, in order to accelerate research developments in 5G technology, supported by a public funding of around €700 million through the Horizon 2020 Programme, and the same amount from the private side. Furthermore, the telecommunications industry will invest five to ten‐times this amount in 5G deployments outside the partnership [14]. The EU industry is set to complement this investment to more than €3 billion. Moreover, the European Commission adopted an Action Plan [2] for a coordinated 5G deployment across all EU member states, targeting early network introduction by 2018, and moving towards commercial large scale introduction by the end of 2020. In order to support this timeline, spectrum should be made available in the 5G pioneer bands [3]: at 700 MHz, within 3.4‐3.8 GHz, and at 26 GHz. Corresponding activities at national level have been initiated, e.g., in Germany [15] and in the UK [16].

In the US, the Federal Communications Commission (FCC) adopted a Report and Order [12] with new rules to enable rapid development and deployment of next generation 5G technologies and services in spectrum bands above 24 GHz. These new rules open up nearly 11 GHz for flexible, mobile and fixed wireless broadband, comprising 3.85 GHz of licensed spectrum in the bands 27.5‐28.35 GHz, 37‐38.6 GHz, 38.6‐40 GHz, and 7 GHz of unlicensed spectrum at 64‐71 GHz. In addition, a government research initiative has been launched [17], including an $85 million investment in advanced wireless testing platforms by a public‐private effort, and an additional $350 million over the next seven years in academic research that can utilize these testing platforms. An auction for spectrum in the 600 MHz band in March 2017 resulted in 70 MHz for licensed use and 14 MHz for wireless microphones and unlicensed use [18], to be freed from television usage by early 2020.

In South Korea, a pilot 5G mobile service is planned for the 2018 Winter Olympics in Pyeongchang in February 2018, and the rollout of 5G commercial services is foreseen for 2019 [19]. In Japan, commercial 5G networks are expected by 2020 in time for the Summer Olympics in Tokyo, or possibly even for the rugby world cup in 2019 [11]. In China, 5G trials are scheduled in two phases, first technology trials from 2015 to 2018, and second product trials from 2018 to 2020, with 5G commercial deployment envisaged by end of 2020 [20]. Note that further details on 5G pilots and early commercial 5G deployments in the different geographic regions are provided in Chapter 17.

3.5.1 Propagation Challenges

The system coverage of wireless networks becomes worse at higher carrier frequencies due to the increase of propagation loss. While propagation modelling is covered in detail in Chapter 4, the key challenges related to propagation at high carrier frequencies are shortly listed here. Depending on the deployment scenario, different propagation aspects are involved, as for instance captured in [21] and also studied in [22].

The base line propagation loss, as detailed in Section 4.2.1, is derived from the frequency dependent free space loss which is distance‐dependent. For high frequencies, depending on the locations of user terminals and base stations (BSs), further propagation components are to be considered in addition, leading to more severe propagation challenges.

In the scenario where user terminals are located indoors and served by outdoor macro BSs, for instance, the building penetration loss, which is typically dependent on the type of material of the outer building walls, is the main challenge, see Section 4.2.3. The incident angle of the main antenna beam to the building entry is another important component which is usually modeled as angular loss. Since macro BSs are usually placed on the rooftop of a building, the main radiation is above or around the edge of the building, causing additional loss known as diffraction loss, see Section 4.3.3. Inside the building, internal wall losses may occur. In addition to these losses related to environment and deployment, also the human body loss may need to be considered, depending on the position of the user device. When user terminals are located outdoors, e.g., in streets, the coverage is not affected by building loss, angular loss, and indoor wall loss. However, due to the placement of macro BSs on rooftops, the coverage is still affected by the diffraction loss and the body loss. In the scenario where indoor users are served by indoor BSs, free space loss, internal wall loss and body loss are affecting the coverage.

3.5.2 Beamforming and 5G Mobile Coverage

With the increase of the carrier frequency, the physical size of one antenna element can be reduced due to decreasing wavelength. Therefore, by keeping the antenna size, the number of antenna elements can be increased, resulting in higher beamforming gain. However, as the propagation loss also increases over the frequency, it is not obvious that the beamforming gain is sufficient for compensating the propagation loss.

In theory, the relation between the power gain G of an antenna, the effective antenna area A, and the wave length λ is as follows [23]:

(3‐1)

Since the wave length is inversely proportional to the carrier frequency, the antenna gain in dBi grows over the frequency f according to the following formula when A is fixed:

(3‐2)

(3‐3)

where c is the speed of light

For Equation (3‐2), the antenna gain increases logarithmically with the carrier frequency when the antenna size is kept unchanged. However, the key frequency‐dependent propagation components in the outdoor to indoor scenarios might increase more rapidly over frequency than the antenna gain. In [21], it is indicated that the building penetration loss in decibel (dB) increases linearly over frequency. For instance, the loss of concrete wall material is given by 5 + 4f (dB).

In reality, the antenna gain at high frequencies might be more limited for a number of reasons. First of all, the beamforming gain might not be continuously increasing over frequency due to hardware limitations or antenna design. Particularly, this would apply for hand‐held devices. Therefore, there will be an upper limit on the number of antenna elements. For instance, it is currently assumed in 3GPP that each TRP is capable to have up to 256 elements in the 30 GHz range, and up to 1024 elements in the 70 GHz range [24]. However, the number of elements may be limited to 32 elements for the UE antenna. In high mobility scenarios, the use of antenna beams smaller than the angular extent of the intended coverage area requires beam steering and tracking functionalities. Even if channel state information (CSI) is available at the transmitter (Tx) and the receiver (Rx) for optimal beamforming, the optimal beam pointing to a mobile user becomes increasingly challenging due to smaller beam widths in higher frequency ranges. Furthermore, regulatory limits for transmission power and electro‐magnetic field (EMF) exposure are to be taken into account.

3.5.3 Analysis of Deployment Scenarios

In this sub‐section, a rough coverage analysis for carrier frequencies up to 100 GHz is performed. For the downlink coverage, three different deployment scenarios, with node types and user locations leading to different propagation characteristics, are examined: outdoor to indoor (O2I), outdoor to outdoor (O2O), and indoor to indoor (I2I).

As mentioned in the previous subsection, there is still uncertainty about the hardware capabilities which might evolve over time. Therefore, analyzing the coverage at high frequencies is a challenging task. Especially, the realizable beamforming gain and the power amplifier efficiency are not easily predictable. Thus, a required system gain is defined as the required average transmit power plus the Tx/Rx beamforming gain in order to reach a certain performance target, and the system gain is estimated over frequencies. Although this metric may not provide the coverage feasibility directly, because the realizable transmission power and the Tx/Rx beamforming gains are still needed, it gives good guidance on the coverage sensitivity at high carrier frequencies to achieve a specific target data rate. In principle, one can conclude that the higher the value of the required system gain, the more difficult it is to achieve the intended target.

In Figure 3‐3, the system gain is shown as a function of the carrier frequency for different deployment scenarios. In this example analysis, a performance target of 50 Mbps was chosen by assuming a channel bandwidth of 100 MHz and a 2x2 MIMO stream at a given thermal noise power with a noise figure of 5 dB. Two different building types are assumed in the O2I scenario: a modern building type consisting of 70% infrared rejection glass windows and of 30% concrete wall, which is similar to a modern building with coated windows, and a classical building type with 30% of standard windows and 70% of concrete, which represents classical buildings in Western Europe. For the O2I and the O2O non‐line‐of‐sight (NLoS) case, a cell edge user located 100 m away from a macro BS is considered. Users in the I2I scenario are assumed to be 10 m away from an indoor BS.

The results show that the O2I scenario is the most challenging one, and most sensitive with regard to a change of the carrier frequency. This indicates that the spectrum up to 30 GHz appears suitable for O2I coverage. In addition, there is a very large variance on the O2I coverage feasibility due to the variety of building types and materials. Compared to the O2I case, the O2O and I2I scenarios have more relaxed requirements on the transmit power and beamforming capability so that carrier frequencies above 30 GHz could be still suitable. However, the beamforming capability is essential to compensate the propagation loss at high frequencies.

3.5.4 Coexistence of 5G Systems and Fixed Service Links

The fixed service is heavily used in some of the 5G candidate frequency bands above 6 GHz. Therefore, the coexistence of rooftop 5G macro‐cell systems and fixed links operating in adjacent channels is to be investigated.

In Table 3‐3, the aggregate adjacent channel interference at the fixed service receiver is shown [22], in dependence of the dish size of the fixed link receiver and the traffic load level within the 5G system. These results are based on system‐level coexistence evaluations in a realistic three‐dimensional (3D) dense urban city with a random height of buildings. In the simulations, the macro BSs and the fixed service link stations were placed on rooftop. A 3D ray‐tracing‐based propagation model was assumed for explicitly modeling diffraction and reflection. In addition, frequency dependent building penetration and wall loss were included. The antenna used in the study assumes UE specific beamforming such that the BS adjusts the beam direction towards a specific UE in a scheduling instance.

The 99th percentile of the aggregate adjacent channel interference in dBm/MHz received at the fixed service receiver is selected as the worst case. The estimated adjacent channel interference level is lower than the typical thermal noise level. In addition, an increase in the dish size of the fixed service station decreases the interference level further due to the higher directivity of the antenna, allowing for a better selection of the wanted signal. The simulations were performed at a carrier frequency of 15 GHz. At higher frequencies, the potential for interference would be less, as the implementation of smaller beams creates more spatial separation in the 5G system, thus enlarging coexistence feasibilities.

3.5.5 Coexistence under License‐exempt Operation

5G is designed to operate in different spectrum usage scenarios (see Section 3.2.1), including the “unlicensed user mode”. Listen‐before‐talk (LBT) is one well‐known mitigation technique to enable the coexistence of wireless systems under license‐exempt operation [25]. This section investigates the usage of LBT and its impact when high gain beamforming is used at high carrier frequencies.

Figure 3‐4 illustrates the downlink system level performance of street micro cells in a dense urban deployment scenario, with and without applying LBT. Two networks, each with four access points (APs) and forty UEs, are considered. The APs are wall mounted and the UEs are distributed randomly outdoor in the streets. In case that beamforming is applied, it is implemented with 100 antenna elements at the BS.

It can be concluded that in general higher system load results in lower throughput per user, but generates an increase of the overall system throughput which is measured by aggregating the traffic from multiple users in the whole system. If beamforming is not applied, it can be observed that a better performance can be achieved with LBT. However, when beamforming is implemented, the impact of LBT becomes marginal. This implies that there is the potential of using high gain beamforming as a mitigation technique for inter‐network spectrum sharing at high frequencies.

3.6.1 Evolutions in Dynamic Spectrum Management

Accurate cooperation between MNOs and national regulatory authorities (NRAs) in terms of effective management of RAN and spectrum resources will be possible only in the presence of a stable, dedicated spectrum management system (SMS) being able to efficiently manage the utilization of spectrum resources available under diverse authorization schemes (e.g., exclusively licensed, LSA, license‐exempt, see Section 3.2.1), i.e. to coordinate and control the realization of agreements, and to enforce the execution of decisions made. In consequence, such a SMS has to provide numerous functionalities to different stakeholders. On one hand, it should be fully automated, realized fully in a software manner, allowing for accurate and real‐time access to shared resources. In order to achieve this goal, a set of standardized interfaces and protocols have to be defined and incorporated into the overall wireless network architecture. On the other hand, rules and policies on how the spectrum and other resources are managed might be modified in various ways, and these modifications can be implemented into the system either automatically or manually by qualified personnel. For example, new policies on spectrum sharing may be provided by the NRA in form of new regulations, or may be the result of mutual agreements between MNOs cooperating in LSA mode. Naturally, the features of a SMS have to be supplemented by effective access to the storage functionalities. The rules for spectrum management, for example those defining the way how operators share spectrum in LSA mode or how the NRA defines the spectrum access policies, have to be effectively stored and should be easily accessible for authorized users of the SMS. In consequence, the presence of dedicated databases or other forms of big data processing and management solutions have to be implemented as well.

Two key solutions for vertical spectrum sharing already exist: an architecture based on the LSA approach, and a three‐tier sharing model controlled by a dedicated Spectrum Access System (SAS). The first one is implemented in Europe by a dedicated standard for the band 2.3‐2.4 GHz [26], and the second one is promoted in the USA for the dynamic management in the 3.5 GHz band [27]. Both systems comprise of a central coordination entity equipped with some intelligence for proper decision making, by using dedicated protocols with a set of messages to be exchanged with the stakeholders. Moreover, the central coordination entity is able to query ancillary databases which are used for storing context information about the ambient environment. One may notice that these two solutions are not well advanced in terms of opportunities for its users, in particular as they are limited to certain frequency bands and to certain rules of spectrum sharing. Thus, in order to move a step forward towards effective dynamic spectrum management, the application of radio‐environment maps (REM) are proposed in the research community.

REMs, also referred to as geolocation databases (GLDB), are considered as an advanced and dedicated tool for effective storage and management of available rich context information, which may be helpful for efficient flexible spectrum access [28]. The REM approach has been considered as a real and pragmatic solution for the problem of unreliable spectrum sensing algorithms which were developed mainly in the context of cognitive radio. REMs are applied as an entity for guaranteeing reliable access to various types of gathered information, and in consequence, for facilitating flexible spectrum access. Most frequently, the databases are considered in the context of storage of interference maps between two radio systems operating in a certain geographical area. Moreover, these structures can store information about the location of wireless transceivers, their transmit parameters and coverage area. Fast and reliable access to such accurate information will enable better spectrum management and thus lead to more efficient spectrum utilization. In a broader sense, REM databases can contain not only interference or coverage information, but may be used also for keeping typical traffic distribution of the mobile users, dedicated and unique, yet anonymized history maps associated with each user. In such a map the traffic distribution over time and space may be provided. In order to realize sophisticated spectrum management, a dedicated REM management system is needed, equipped with an entity responsible for database queries and management, and a dedicated engine for the analysis of the data on users and inference. Thus, a REM based spectrum management system seems to be highly similar to the generic concept of the SMS considered earlier.

The future SMS has to be discussed also in the context of effective coexistence between cellular mobile and Wireless Local Area Network (WLAN) deployments. Mobile data offloading from cellular to other types of networks is one solution for effective management of high user‐data traffic. MNOs and hardware manufacturers are widely considering coexistence and even cooperation between technologies operating in licensed and in license‐exempt spectrum, for example the License Assisted Access (LAA) scheme using carrier aggregation to combine LTE in the license‐exempt 5 GHz spectrum with LTE in licensed bands. With this aggregation, where control and signaling information is transmitted in the licensed spectrum, higher user data rates can be supported, while the user experience remains seamless and reliable. The convergence of cellular and non‐cellular networks is supported by technical developments in both research and standardization communities. In 3GPP‐based networks, the presence of the Access Network Discovery and Selection Function (ANDSF) entity, as well as Local IP Access (LIPA) or Selected IP Traffic Offloading (SIPTO), are the exemplary steps toward this direction. In the IEEE community, the introduction of the Wi‐Fi Certified Passpoint™ and the Hotspot 2.0 concept, together with advanced databases and the dedicated Access Network Query Protocol (ANQP), pave the way for tighter cooperation between cellular and non‐cellular networks. Further information can be found in [29] and [30]. The discussion above leads to the following conclusions: First, there are already technologies and technical solutions available for an effective management of traffic between these two types of networks. Second, there is a strong need for a RAT‐independent SMS. The latter should be treated as a solution for the effective organization of various types of spectrum resources and RATs. Again, the application of such a RAT‐agnostic SMS requires the presence of a dedicated inference engine, standardized protocols and interfaces for message exchange, as well as accurate databases.

Finally, one may observe that in fact most of the contemporary wireless networks are based on IP‐centric solutions. Thus, MNOs may benefit from the virtualization of their resources. Indeed, SDN together with the application of NFV opens the door for a fully software‐controlled management and optimization of networks. In such a context, the data of mobile users connected to the network may be managed by means of virtualized entities. One can even claim that if the service‐level agreements (SLAs) between end‐users and service providers are fulfilled, the way how the data is transported, i.e., over cellular or non‐cellular connection, is of secondary importance. As the virtualization techniques are now widely applied to various types of networks, one may again assume that there is a need for an advanced, flexible spectrum management system that will be able – in a wide sense – to coordinate and control the usage of spectrum resources from a common pool. A comprehensive discussion on wireless network virtualization may be found for instance in [31].

3.6.2 Functional Spectrum Management Architecture

Advanced spectrum management systems are already under consideration in different research projects [6], [32]. A possible functional architecture of a holistic SMS is illustrated in Figure 3‐5, embracing the key functionalities required from a regulatory (NRA domain) as well as from an operational (MNO domain) point of view.

SMS functionalities within the NRA domain:

• Regulatory Spectrum Coordination: From NRA perspective, the management and coordination of spectrum requires the existence of a central coordination point which is permitted to perform any necessary action to guarantee proper execution of NRA rules. This entity has to handle also the communication with the MNO domains.
• System Management: As already mentioned, the SMS should be able to incorporate any modifications of the existing (applied) spectrum management policies or to include new ones. These policies may be prepared in human‐ and/or machine understandable form, and be provided either in an automated way, or implemented by qualified personnel, while the system management is providing functions to perform operation administration, and maintenance tasks.
• Spectrum User Authorization: From the NRA perspective it is evident that the SMS system should provide functionality for secure user authorization. Thus, this entity is containing MNOs licensing and registration information.
• Data storage: Information of interest from the perspective of NRA (such as spectrum resources, protection and usage rules, etc.) has to be stored in an efficient form. Dedicated repositories and databases shall be available in the SMS, which may create a fragment of the overall REM‐based structure.
• Monitoring functionality (including reporting and information entry): In order to realize fair and effective management of the spectrum, but also to avoid violation of policies provided by the stakeholders, advanced monitoring functionalities have to be available to the NRA. These modules will provide updates to the Regulatory Spectrum Coordination entity, which in turn will provide updates to the databases.

SMS functionalities within the MNO domain:

• Spectrum Assignment Coordination: This entity plays a central coordination role. It is responsible for spectrum assignment (radio resource management) by utilizing information from various data bases and spectrum sharing enablers if applicable. It also coordinates the mutual dependencies between operators (defined and stored in the databases), and communicates with the Regulatory Spectrum Coordination entity in the NRA domain.
• Dedicated interfaces due to ownership issues: There will be a dedicated client of the SMS system per each MNO due to the fact that each one needs to manage sensitive information (e.g., data related to customers, business etc.). However, in a virtualized scenario, such a spectrum management application (client) needs to communicate with other involved stakeholders to exchange necessary data for proper spectrum management. Thus, both dedicated interfaces between the involved MNOs and between MNOs and the NRA, and advanced inter‐operator functionalities have to be incorporated in the system.
• Policy analysis: some of the policies may be generic (e.g., those provided by the NRA and defining the rules of thumb for a certain services, frequencies or geographical areas), whereas others may be more specific (such as those defined by the mutual agreements between two MNOs). As the SMS is assumed to be RAT‐agnostic (i.e., various RATs may be used simultaneously), there is a risk that these policies may be somehow mutually depended, or even incoherent. In order to avoid any problems related to this aspect, the SMS should be equipped with a dedicated reasoning and inference engine being able to apply advanced inference and machine learning algorithms.
• Access to Databases: Information received from the NRA but also from other MNOs, or gathered by dedicated monitoring modules, have to be stored in the databases. These databases may be part of broader REM‐based structure. In case of network virtualization, these databases will include for example service‐specific requirements or rules related to spectrum usage. Three types of repositories are included in Figure 3‐5: Service‐Specific Spectrum Requirements, Spectrum Resource Storage, and Spectrum Usage Rules.

It is to be noted that the above split of functionalities does not provide any conclusion on how the SMS is implemented. In particular, although the spectrum coordinator entities are centralized in a logical way, in practice there may be a number of coordination modules, each responsible for a certain geographical area. Moreover, a hierarchical deployment scenario may be considered, where the long‐term analysis (such as reasoning for potential policy updates, analysis of the traffic, application of LSA rules etc.) may be implemented in the core network, and short term decisions (such as scheduling, vertical and horizontal handover management, etc.) will be placed in the RAN.

References

1. 1 ITU Radio Regulations, Edition of 2016
2. 2 COMMUNICATION FROM THE COMMISSION TO THE EUROPEAN PARLIAMENT, THE COUNCIL, THE EUROPEAN ECONOMIC AND SOCIAL COMMITTEE AND THE COMMITTEE OF THE REGIONS, COM(2016) 588, “5G for Europe: An Action Plan”, September 2016
3. 3 RADIO SPECTRUM POLICY GROUP Opinion on spectrum related aspects for next‐generation wireless systems (5G), “Strategic Roadmap Towards 5G for Europe“, November 2016
4. 4 GSMA, 5G Spectrum, Public Policy Position, November 2016
5. 5 5G PPP METIS‐II (Mobile and wireless communications Enablers for Twenty‐Twenty Information Society II) project, see https://5g‐ppp.eu/metis‐ii/
6. 6 5G PPP METIS‐II project, Deliverable D3.2, “Enablers to secure sufficient access to adequate spectrum for 5G”, June 2017
7. 7 RECOMMENDATION ITU‐R M. 2083‐0, “IMT Vision – Framework and overall objectives of the future development of IMT for 2020 and beyond”, September 2015
8. 8 Report ITU‐R M.2290‐0, “Future spectrum requirements estimate for terrestrial IMT”, December 2013
9. 9 3GPP TSG RAN WG1 Meeting #88bis, Document R1‐1705579, “OTA calibration for multi‐TRP transmission”, April 2017
10. 10 Cisco, White Paper, “Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2016–2021”, February 2017
11. 11 GSA Executive Report from Ericsson, Intel, Huawei, Nokia and Qualcomm, “The case for new 5G spectrum”, November 2016
12. 12 Federal Communications Commission, “REPORT AND ORDER AND FURTHER NOTICE OF PROPOSED RULEMAKING”, July 2016
13. 13 ERC REPORT 25, “THE EUROPEAN TABLE OF FREQUENCY ALLOCATIONS AND APPLICATIONS IN THE FREQUENCY RANGE 8.3 kHz to 3000 GHz (ECA TABLE)”, June 2016
14. 14 European Commission, “5G Infrastructure PPP: The next generation of communication networks will be ‘Made in EU’”, see: http://ec.europa.eu/research/press/2013/pdf/ppp/5g_factsheet.pdf
15. 15 Bundesnetzagentur, “Points of Orientation for the provision of spectrum for the rollout of digital infrastructures”, December 2016
16. 16 Ofcom, “Update on 5G spectrum in the UK”, February 2017
17. 17 The White House, “Fact Sheet: Administration Announces an Advanced Wireless Research Initiative, Building on President’s Legacy of Forward‐Leaning Broadband Policy”, July 2016, see: https://obamawhitehouse.archives.gov/the‐press‐office/2016/07/15/fact‐sheet‐administration‐announces‐advanced‐wireless‐research
18. 18 FCC, “Broadcast Incentive Auction and Post‐Auction Transition”, May 2017, see: https://www.fcc.gov/about‐fcc/fcc‐initiatives/incentive‐auctions
19. 19 Ministry of Science and ICT, Press Release: “Ministry Highlights 5G During G20 Digital Ministerial Meeting”, April 2017, see: http://english.msip.go.kr/english/msipContents/contents.do?mId=Mjc0
20. 20 Ministry of Industry and Information Technology of the People’s Republic of China, “5G Progress in China”, November 2016, see: https://5g‐ppp.eu/wp‐content/uploads/2016/11/Opening‐1_Qian‐Hang.pdf
21. 21 3GPP TR 38.900, “Study on channel model for frequency spectrum above 6 GHz”, V14.2.0, December 2016
22. 22 5G PPP METIS‐II project, Deliverable D3.1, “5G spectrum scenarios, requirements and technical aspects for bands above 6 GHz”, May 2016
23. 23 C. A. Balanis, “Antenna Theory: Analysis and Design”, 4th Edition, Wiley, 2016
24. 24 3GPP TR 38.802, “Study on New Radio Access Technology Physical Layer Aspects”, V1.1.0, January 2017
25. 25 Draft ETSI EN 302 567, “Multiple‐Gigabit/s radio equipment operating in the 60 GHz band; Harmonised Standard covering the essential requirements of article 3.2 of Directive 2014/53/EU”, V2.0.22, December 2016
26. 26 ETSI TS 103 235, “Reconfigurable Radio Systems (RRS); System architecture and high level procedures for operation of Licensed Shared Access (LSA) in the 2300 MHz ‐ 2400 MHz band”, V1.1.1, October 2015
27. 27 President’s Council of Advisors on Science and Technology (PCAST) Report, “Realizing the Full Potential of Government‐Held Spectrum to Spur Economic Growth”, July 2012
28. 28 V. Atanasovski et al., “Constructing radio environment maps with heterogeneous spectrum sensors”, 2011 IEEE International Symposium on Dynamic Spectrum Access Networks (DySPAN), May 2011
29. 29 3GPP TR 36.889, “Study of Licensed‐Assisted Access to Unlicensed spectrum”, V13.0.0, June 2015
30. 30 Dino Flore, Chairman of 3GPP TSG‐RAN, “3GPP & unlicensed spectrum”, IEEE 802 Interim Session, Atlanta, USA, January 2015
31. 31 C. Liang and F. R. Yu, “Wireless Network Virtualization: A Survey, Some Research Issues and Challenges”, IEEE Communications Surveys & Tutorials, Vol. 17, Issue 1, pp. 358–380, Firstquarter 2015
32. 32 5G PPP COHERENT project, Deliverable D4.1, “Report on enhanced LSA, intra‐operator spectrum‐sharing and micro‐area spectrum sharing”, June 2016
33. 33 H2020 TWEETHER project, see: https://tweether.eu/
34. 34 H.‐J. Song, K. Ajito, A. Wakatsuki, Y. Muramoto, N. Kukutsu, Y. Kado, T. Nagatsuma, “Terahertz wireless communication link at 300 GHz”, IEEE Topical Meeting on Microwave Photonics, October 2010, see: https://www.researchgate.net/publication/224204378_Terahertz_wireless_communication_link_at_300_GHz