2.4.1 Licensed Spectrum

Dedicated and licensed spectrum will continue to be preferred for many 5G key use cases. Licensed spectrum ensures a stable framework for investment with guaranteed coverage and QoS. This can be achieved as with a given licensed spectrum block, the licensee has the right to control all radio communication. The licensee, who is in the case of mobile communication, the mobile network operator, can thus perform accurate network planning including linking budget calculations, interference and system capacity modeling without the risk of interference from other users using the same spectrum.

The co‐channel interference is caused by the operator's own mobile network operations, which can be controlled and mitigated by network algorithms and interference cancelation receivers on both the network and on the mobile side. Interference from neighboring channels is controlled by adjacent channel leakage requirements, set for all transmitters operating at the mobile spectrum. Additionally, for receivers' adjacent channel rejection and out of band blocking requirements the operator can define the minimum requirement for reducing adjacent channel and out of band interference. These requirements are defined in 3rd Generation Partnership Project (3GPP) Radio Access Network (RAN) Working Group (WG) 4, for each supported band to ensure that all mobile devices and BS support these minimum requirements, so that the regulatory requirements are met, and interference conditions in each carrier can be ensured. Therefore, licensed spectrum also remains the primary spectrum asset for mobile operators for 5G deployments.

Typically, as was the case with 2G, 3G and 4G deployments, these licenses are sold in spectrum auctions in a country or assigned to mobile operators under specific terms. Additionally, depending on the country the spectrum auction may set rules how the spectrum owner must utilize the spectrum, e.g. the spectrum may need to be taken into use within a specific time after the license has been granted and/or the deployment must meet certain coverage requirements (covering certain percentage of the country or population).

Mobile operators have been investing on spectrum since the addressable market has been significant, and utilization of auctioned spectrum is mainly market driven, i.e. operators can deploy their networks in areas where it is justified from a business perspective. Additionally, one should not underestimate the fact that by investing into the spectrum auction, existing mobile operators can avoid new competitors accessing the market.

In some smaller European countries like Finland, specific terms for auction has been applied that limits the amount of spectrum that each bidder can obtain, but also set requirements to when spectrum needs to be taken into use and to provide significant network coverage. An example of such terms is the frequency band allocation for telecom operating licenses for the spectrum 703–733 MHz and 758–788 MHz in Finland, which was defined as follows (see [3]): “License must be built so as to cover 99% of the population of mainland Finland within three years of the start of the license period. The coverage requirement should be structured so as to ensure reasonable indoor coverage within the coverage area. ‘Reasonable indoor coverage’ means that the telecom operator's services must be available without additional cost to users in normal circumstances of use in users' permanent residences or enterprises' places of business.”

In addition, the rule defined that “No more than two 2 × 5 MHz frequency pairs will be allocated to any individual enterprise or organization.” However, in coverage calculation, the broadband mobile communications networks previously built by the license holder using the 2.6 GHz, 1800 MHz and 800 MHz spectra are considered [3]. At the end, existing mobile operators bought the maximum frequency amount, almost with starting bid of 11 million Euros per frequency block (only TeliaSonera paid 11 330 000 Euros on other frequency blocks) as the cost of coverage build was prohibitive for new entrants and the existing operators could not compete against each other on the amount of spectrum [4].

However, such licensing rules lead to a situation where mobile telecommunication clearly flourishes in Finland, as mobile subscriptions prices are inexpensive, and utilization of mobile data is very high. Figure 2.1 presents data usage per mobile broadband subscription on average for OECD (Organisation for Economic Co‐operation and Development) countries (for details see [5]). It can be observed that in Finland the average data usage was around 15.5 Gigabytes (GB) per month, which is over five times more than the OECD average and almost 9 times more than in Germany where the monthly costs of a mobile subscription are relatively high.

It is apparent that licensing rules have an impact on the utilization of the spectrum, thus regulators need to balance between several terms such as ensuring competition, auction income, spectrum utilization, license period and impact of auction prices to mobile subscription prices, as well as to identify enough spectrum to be licensed. Both license‐exempt (unlicensed) spectrum and new regulation approaches can provide additional tools in this process.

2.4.2 License‐Exempt Spectrum

Instead of licensed spectrum, where radio operation is controlled by the licensee, the radio operator in license‐exempt spectrum (Table 2.3) is free if operation fulfills the regulation requirements. These requirements depend on the frequency band and are country specific. The best harmonization has been achieved for the 2.4 GHz Industrial Scientific and Medical (ISM) band which has only few global variations. Below 1 GHz frequency, there is the 915 MHz band in the US and other countries following the Federal Communications Commission (FCC) regulation in Canada, South Korea, and Australia. This frequency band is not available in Europe where frequency blocks between 863 and 870 MHz have been allocated for license‐exempt use, however, this band is not completely harmonized inside Europe and thus it has variations on spectrum range between European countries. Spectrum below 1 GHz is used for the growing number of Internet of Things (IoT) devices and low power wide area networks (LPWA) but have not been addressed by 3GPP in previous license‐exempt spectrum solutions as available spectrum and transmission powers are limited. It needs to be seen whether any of these bands will be addressed by 3GPP, e.g. for IoT purposes.

As has been said, the best harmonization on unlicensed license‐exempt has been obtained for the 2.4 GHz band. Thus, it is tremendously utilized by Wi‐Fi and Bluetooth. This spectrum block is also being more and more utilized by different IoT short range technologies, either with Bluetooth Low Energy (BLE) physical layer solution, other standard solutions such as ZigBee, or proprietary radio solutions. Due to the good propagation environment and the high number of available devices supporting this band, it will remain very important for existing and forthcoming Wi‐Fi implementations. However, as interference is increasing in this band, bandwidth demanding applications will and have already moved to the less crowded 5 GHz band.

The 5 GHz band is also the primary band for 5G license‐exempt solutions as it provides a significant amount of spectrum and is optimized for wideband transmissions for mobile broadband use cases. The requirement for wideband transmission is, e.g. defined in the European regulation via Power Spectral Density as shown in Table 2.4. For details see also [9]. Due to spectral power limitations the maximum transmission power can be only achieved, if the transmission bandwidth is 20 MHz, and narrower transmission need to reduce Transmitter Exchange (TX) power to meet the power spectral density limit. This is the case in LTE unlicensed spectrum operation when the eNB (enhamced Node‐B, i.e. a BS) is transmitting the Broadcast Control Channel (BCCH) at 1.4 MHz bandwidth.

In addition to existing spectrum, it is expected that also unlicensed‐exempt spectrum will be allocated during WRC‐19 process or as part of it. The NR operation on unlicensed spectrum was not part of the first 3GPP NR release (Release 15), but it is planned to be addressed as part of Release 16.

2.4.3 New Regulatory Approaches

As discussed above, dedicated and licensed spectrum will continue to be preferred for many key use cases of 5G especially where coverage and QoS needs to be guaranteed. Additionally, license‐exempt spectrum can also play an important role for 5G. However, novel regulatory approaches and tools such as Licensed Shared Access (LSA) and Spectrum Access System (SAS) are expected to complement the existing dedicated and licensed spectrum access regimes. Future access to new spectrum for 5G in the 4–6 GHz range may depend on sharing possibilities with incumbent users in many regions where re‐purposing of spectrum is not possible. This can provide means to establish dedicated or private networks for certain use cases in a limited area or region.

These innovative ways of allocating even licensed spectrum to users (not necessarily mobile operators) are under discussion in many countries. Low latency, high reliability use cases like collaborative robots and industry automation make it necessary to allow usage of licensed spectrum in local or regional areas for specific applications. In Germany, an auction of frequencies in the range of 3.6 GHz is planned for beginning of 2019. In parallel, frequencies in the range of 3.700–3.800 MHz and 26 GHz are made available for local or regional use based on request (i.e. no auction is planned). This allows deployment of local networks in factories, harbors, airports, campuses, cities and hot‐spots like a stadium. Local frequencies need to be used within one year after allocation, otherwise the user must return the license to the regulator. Use of such local frequencies is also planned in other European countries like the Netherlands.

Additionally, operation in some of mmWave bands may initially be amenable to spectrum and network infrastructure sharing arrangements between operators to avoid installation of multiple network equipment (especially antennas) at each site and therefore reducing deployment costs. In general, 5G systems need to support different spectrum authorization modes and sharing scenarios in different frequency bands.

As illustrated in Figure 2.2, four different user modes can be defined under which 5G radio access systems are expected to operate, namely the “service dedicated user mode,” the “exclusive user mode,” the “LSA user mode” and the “unlicensed user mode” (see [12]). The use of radio 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. The relationship between user modes and authorization schemes is visible in the upper part of Figure 2.2, named “regulatory framework domain.”

Spectrum usage rights awarded by “individual authorization” are exclusive at a given location and/or time. The “service dedicated user mode” refers to spectrum designated to services other than Mobile/Fixed Communication Network (MFCN) operation, which are indented to be integrated into the 5G eco system, e.g. Intelligent Traffic Systems (ITS) or Public Protection and Disaster Relief (PPDR) services. This spectrum is only used for dedicated services and applications. Spectrum designated to MFCN falls into the “exclusive user mode.” In the “LSA user mode” a non‐MFCN license holder (incumbent) would share spectrum access rights with one or more mobile network operators (LSA licensee), which can use the spectrum under defined conditions, subject to individual agreement and permission by the relevant regulatory authority.

The user modes can occur in their basic form (continuous lines), or as evolution of current approaches in the form of “limited spectrum pool” or “mutual renting” (dashed lines), see Figure 2.2. Limited spectrum pool is used in spectrum usage scenarios where a limited number of known operators obtain authorizations to access a spectrum band dynamically. It is envisioned that mutual agreements between licensees are such that the long‐term share of an individual operator has a predictable minimum value.

Mutual renting allows an operator to rent at least part of its licensed spectrum resources to another operator, based on mutually agreed rules. While the spectrum ownership stays unchanged, the rules may define spectrum usage restrictions and spectrum owner protections. Mutual renting can provide both static (i.e. like exclusive use), and/or dynamic shared spectrum.

Depending on the duration (static or dynamic) of the spectrum access, the spectrum usage scenarios “limited spectrum pool,” “mutual renting” , and “vertical sharing” are considered as exclusive use or shared use.

In the “unlicensed user mode” spectrum access and usage rights are granted by general authorization, i.e. without an individual license, but subject to certain technical restrictions or conditions like limited transmit power or functional features like duty cycle or listen‐before‐talk as required in many license‐exempt bands. In this mode, users cannot claim protection and may receive interference from other users.

Spectrum sharing between systems of different priority, e.g. if incumbent users must be protected in the “LSA user mode,” is referred to as “vertical sharing,” and sharing between systems of equal priority is called “horizontal sharing.” For example, 5 GHz WLAN systems need to ensure protection of the incumbent radar systems (vertical sharing), and coexistence with other WLAN systems (horizontal sharing).

To achieve high spectrum usage efficiency, 5G systems may preferably support all spectrum usage scenarios shown in Figure 2.2, noting that several scenarios may occur simultaneously.

2.5.1 Pathloss

A first required evaluation in system design is to estimate pathloss with different receiver distances (Figure 2.3). For details see also [16]. A good estimation of the pathloss can provide a very intuitive view to the system as part of link budged calculations. With given transmission powers and antenna configurations achievable data rates can be estimated on a given link distance. The benefit of such calculations is that these results can be obtained without any detailed link or system simulations. Additionally, the pathloss models are used in system simulations as well as in network planning tools when no detailed environment information is available.

The 3GPP channel models cover several different environments, each having different pathloss models separately for both Line of Sight (LOS) and Non‐Line of Sight (NLOS) conditions. Supported environments are Rural Macro (RMa), Urban Macro (UMa), Urban Micro Street Canyon (UMi – SC) and Indoor Hotspot (InH). For example, the indoor NLOS pathloss is defined by following functions with additional random loss based on normal distribution N(0,σ²_P) probability function with an 8 dB standard deviation:

where PL_InH‐LOS is the pathloss for LOS conditions and defined as

and PL′_InH‐NLOS is defined as

The function shown above assumes that both transmitter and receiver are inside the building and there is a signal path from transmitter to receiver so that penetration through internal walls is not needed. By using similar kind of formulas, the pathloss for other environments can be calculated.

When considering an outdoor BS and indoor mobile terminals the total pathloss is the sum of outdoor pathloss PL_b, indoor pathloss PL_in and additional wall penetration loss PL_tw as shown in the following formula:

The PL_tw is defined by the formula:

where PL_npi is a constant value of 5 dB currently used in [16], and p_i is the proportion of the i‐th material of the wall. The carrier frequency dependent values for each material L_{materila_i} is given in Table 2.5. This model allows e.g. calculating wall penetration loss for walls with certain percentage of concrete and standard multi‐panel glass.

2.5.2 Multipath Propagation

After understanding pathloss to get a high‐level view of the achievable link budget, it is essential for the system design to understand multipath propagation of the transmitted signal in different environments. The transmitted signal typically propagates through multiple paths, each path introducing different amplitude and phase to the received signal, causing frequency selective channel, i.e. frequency selective fading. This can be further used in frequency selective scheduling in Orthogonal Frequency‐Division Multiplexing (OFDM) based systems where only strong frequency components, OFDM subcarriers, are used for transmitting data to the intended receiver. Fading frequency parts are left unused and assigned to other receivers, which have different multipath channel conditions. Thus, fading occurs to different OFDM subcarriers. Furthermore, in system design the frequency selectivity can be considered when defining reference symbols for channel estimation processes, so that the receiver's equalizer can compensate fading and phase distortion of the different OFDM subcarriers.

Additionally, the delay spread of these multipath signals is an important design criterion for physical layer numerology of the OFDM signal waveform. The OFDM signal utilizes Fast Fourier Transform (FFT) processing in the receiver, which can recover multipath signal components that are not delayed more than a cyclic prefix (CP) of the OFDM symbol. This enables the receiver to utilize the received energy of all multipath components in the demodulation process. However, multipath signal components that fall outside of a CP introduce inter symbol interference (ISI) from that portion of the symbol falling on the next symbol. To avoid ISI completely, the used CP should be longer than the delay of any multipath component received with any meaningful power. However, as CP is direct overhead to an OFDM signal, it cannot be made extensively long compared to the overall symbol duration, rather the CP length is a compromise between expected delay spread, used symbol length, and acceptable ISI and system overhead.

It is important to recognize that the deciding factor is the relative signal power of these different multipath components. Thus, even if there are multipath components that fall outside the CP, the ISI does not immediately destroy the data reception completely, but rather introduces noise floor to the signal that can become dominant in high Signal to Noise Ratio (SNR) conditions and can therefore be the limiting factor to utilize high modulation orders such as 256 QAM or 512 QAM (QAM stands for Quadrature Amplitude Modulation). Thus, longer symbols and lower modulation orders are more tolerant on ISI caused by delay spread, however longer symbols lead to narrower subcarrier spacing, which is more vulnerable for phase noise. The different waveform and OFDM numerology issues are further discussed in Chapter 3.

Finally, the understanding of angle of arrival (AOA) and angle of departure (AOD) properties in different environments allows exploding multi‐antenna systems (see [17]). Multi‐antenna systems with efficient beamforming solutions are essential on high millimeter operating frequencies to compensate high pathloss. Beamforming further increases the relative power of main multipath components, reducing the impact of delay spread and frequency selective fading. Beamforming and beam selection/steering are further discussed in Chapter 3.