Nokia Networks
FutureWorks
Ten key rules rules of 5G depl deploymen oymentt Enabling 1 Tbit/s/km2 in 2030
Nokia Networks white paper Ten key rules of 5G deployment de ployment
Contents
Executive Summary
3
5G system requirement: 1 Tbit/s/km 2
4
5G deployment options
6
5G deployment recommendations
9
Summary and conclusions
Page 2
15
networks.nokia.com
Executive Summary By 2030 there is likely to be as much as 10,000 times more wireless data traic criss-crossing networks than there was in 2010, according to Nokia estimates. The growth will be driven by the use of ultra-high resolution video streaming, the ubiquity of cloud-based applications, entertainment and greater use of high resolution screens at form factors we may not even guess at today. As well as more of the same, we will see new use cases, applications and devices stemming from the powerful trend of the Internet of Things, which will lead to what we call the programmable world. In addition, 5G will provide at least a ten-fold improvement in the user experience compared to 4G in terms of peak data rates and minimal latency. Nokia envisions 5G as being a system that provides a scalable and exible service experience with virtually zero latency, involving gigabits of data when and where it matters.
This white paper outlines the deployment options for 5G to provide the required capacity and end user data rates that will be needed. The ten key recommendations for 5G deployments are: • LTE Advanced can provide the required capacity of tens of Gb/s/km2 for 2020 and beyond. • Approximately 1 GHz of aggregated spectrum to provide the required capacity and cell edge data rates by 2030. • A 5G small cells deployment in 6-30 GHz band (cmWave) with a 500 MHz carrier bandwidth can provide hundreds of Gb/s/km 2 for 2025 and beyond. • A 5G small cells deployment in up to 100 GHz band (mmWave) with 2 GHz carrier bandwidth can provide a Tb/s/km2 for 2030 and beyond. • mmWave can further provide backhaul to the small cells in a mesh conguration with a maximum of two hops. • Very large antenna arrays can be used to eectively compensate for the higher path loss at higher frequency bands. • For both the cmWave and mmWave deployments an inter-site distance of 75-100 m can provide full coverage and satisfy the required capacity, depending on the environment. • A 5G wide area solution is needed to provide the required coverage and cell edge data rates for 2030. • Dedicated indoor small cell deployments are needed to satisfy indoor capacity requirements beyond 2020. • Multi connectivity between LTE Advanced, cmWave, and mmWave signicantly boosts cell edge performance and can lower the required density for small cell deployment. This white paper provides an overview of the 5G requirements and deployment options in the spectrum from 2 GHz up to 100 GHz. The paper concludes with key recommendations for deploying a 5G system from 2020 towards 2030.
Page 3
networks.nokia.com
5G system requirement: 1 Tbit/s/km2 Mobile broadband is the key use case today and is expected to remain one of the key use cases that will set the requirements for 5G. Mobile broadband goes far beyond basic mobile Internet access and covers rich interactive work, media and entertainment applications in the cloud or reality augmentations (both centralized and distributed). Currently, mobile data traic is roughly doubling every year and is expected to continue to grow towards 2030. The strong growth is expected to continue towards 2030. The need for more capacity is just one driver for mobile networks to evolve towards 5G. The full set of key requirements foreseen by Nokia for 5G is shown in Figure 1.
>10 Gbps
Massive Broadband
peak data rates
10 000
10-100
M2M
Smart city cameras
ultra low cost
Massive machine type communication
(Low power) Wide area
x more traff ic
Capacity for everyone
x more devices
10 years on battery
100 Mbps
whenever needed
Sensor NW
Ultra reliability
3D video / 4K screens Work in the cloud
Flexibility VR gaming for the Industry 4.0 unknown
<1 ms
Remote control of robot Mission critical broadcast Autonomous driving
radio latency
Critical machine type communication
# of Devices | Cost | Power Crowd
A trillion of devices with diff erent needs
Ultra-dense GB transferred in an instant
Outdoor Mission-critical wireless control and automation
Figure 1. 5G will enable very diverse use cases with extreme range of requirements The huge amount of traic will need to be carried through all mobile broadband technologies at some point between 2020 and 2030. The need for more capacity will demand more spectrum at higher carrier frequencies. Thus, the 5G system needs to be designed for deployment in new frequency bands as well as coexisting and integrating with other radio access technologies. The growth in mobile data traic will be accompanied by an increase in the number of communication devices. We expect to see ten to one hundred more devices per mobile communications user, ranging from phone, tablet, laptop and smart watches to smart shirts. In addition the number of
Page 4
networks.nokia.com
connected machines and sensors in industry and the public infrastructure is increasing. This trend will continue and 5G will need to accommodate growth in the number of devices. A battery life of 10 years will be needed for machine-type communications (MTC). As the technology evolves, battery life will be improved but this is not the full story. More eicient handling of machine type traic in the 5G system will also be needed, even though a 10-year battery life can already be achieved for MTC with LTE Rel-13/14. The ability to handle very low cost devices must be present across the whole range of 5G frequency bands.
Radio latency less than one millisecond is important to a whole new range of use cases, such as remote control of machines and objects in the cloud or the tactile Internet. Low latency also ensures the system responds quickly, for example fast wake up and dormancy, fast scheduling and fast link reconguration. Lower latencies for the end user will come from higher transmission data rates, but also through appropriate design of the 5G system. The peak data rate of a 5G system needs to be higher than 10 Gbit/s but more importantly, the cell-edge data rate (guaranteed for 95% of the users) should be at least 100 Mbit/s, enabling the mobile Internet to act as a reliable substitute for cable networks wherever needed. Combining data growth and user data rates with a denser subscriber base, we can calculate the total area capacity that 5G will need to support by 2030. This assumes: • Traic per subscriber per day: 30 GByte personalized data • Subscriber density: 100,000 users/km2 • Busy hour traic: 10% of the daily traic. These requirements will necessitate a 5G system that can support ~1 Tbit/s/ km2 in 2030. This paper describes the deployment options and recommends how to deploy a 5G network to cope with this degree of traic density. It considers requirements for cell density, backhaul and spectrum, as well as recommendations for HetNet congurations.
Page 5
networks.nokia.com
5G deployment options LTE-Advanced is today’s favored option for providing mobile broadband for both macro and small cells. Nokia has extensively researched LTE deployment to show that an LTE-based HetNet can cope with a capacity up to a thousand times greater than that common in 2010. To meet capacity needs beyond this gure, small cells using 5G frequency bands need to be deployed with an LTE macro/HetNet overlay. The key challenge for LTE Advanced to provide an excellent end-user experience is to satisfy the demand for cell-edge data rates that will grow to 100 Mbit/s in 2030. This requires a higher bandwidth compared with existing spectrum allocation below 6 GHz. Our analysis shows that up to 2 GHz of spectrum could be used for cellular below 6 GHz. This will be divided among several operators, so for example, four operators would receive only 500 MHz each. To satisfy the 5G requirements for capacity and data rates, new and more advanced 5G systems are needed. Nokia foresees 5G using the full spectrum range, from below 1 GHz to 100 GHz, providing wide area coverage and high capacity in dense areas. While more spectrum below 6 GHz is needed and new promising techniques such as LSA/ASA will increase the use of existing frequencies, there will be an increasing need to unlock new spectrum bands from 6 to 100 GHz for mobile use. This range can be broadly split into two parts, centimeter wave (cmWave) and millimeter wave (mmWave), based on dierent radio propagation characteristics and possible carrier bandwidth. For cmWave frequency bands where moderate system bandwidths such as 500 MHz are used to meet 5G requirements, massive MIMO will increase spectral eiciency -peak, average user and cell edge. This uses a large number of parallel transmit streams (high-rank single user MIMO (SU-MIMO) e.g. 8 streams) as well as multi-user MIMO (MU-MIMO). For mmWave frequency bands where large bandwidths such as 2GHz are used, a large number of antennas can be used to generate narrow beams to mitigate the increased path loss, coupled with low rank SU-MIMO (e.g. 2-4 streams) as well as MU-MIMO. Similar coverage can be achieved at dierent bands ranging from 28-100 GHz, since for the same form factor, larger antenna arrays can be used in high bands to compensate for the path loss dierence between high and low bands.
Page 6
networks.nokia.com
Table 1. System conguration for LTE-Advanced and evaluated 5G systems from 6-100 GHz Parameter
LTE-Advanced
cmWave
mmWave
Frequency band
≤6 GHz
6-30 GHz
30-100 GHz
Carrier bandwidth
100 & 200 MHz
500 MHz
2 GHz
Modulation order
64 QAM
256 QAM
64 QAM
MIMO combination
8x8
8x8
2x2
SU-MIMO rank
8
8
2
MU-MIMO rank
2
2
2
Antenna conguration
10x1 AAS 8 antenna ports MIMO (macro)
Omni directional 4 antenna ports
4x4 AAS 4 sectors 2 antenna ports
Table 1 outlines the key assumption used for three dierent systems LTE-Advanced, cmWave, and mmWave. To make a realistic model and analysis we have focused on the radio wave propagation from 2 GHz to 73 GHz, which has been used throughout this study. Figure 2 shows the line of sight (LoS) path loss comparison for the dierent frequency ranges. The shape of the CDF shows that the path loss components are consistent across the frequency bands. In our 5G deployment analysis, LTE-Advanced is deployed at 2 GHz with up to 100 or 200 MHz bandwidth; the cmWave system is deployed at 10 and 28 GHz with a 500 MHz bandwidth, while the mmWave system is deployed at 28, 39 and 73 GHz with a 2 GHz bandwidth. Ray tracing based on 3D city models was used to generate the path losses used in the case study. The variation in path loss is based on the distance between the device and the eNB. 1 y t i s 0.9 n e d 0.8 e v i 0.7 t a l u 0.6 m u c 0.5 e h t 0.4 n o 0.3 y t i l i n 0.2 b o a i t b c 0.1 n o r u P f 0
2GHz 5.6GHz 10.25GHz 28.5GHz 39.3GHz 73.5GHz
Freq.
50
60
70
80
90 100 Pathloss [dB]
110
120
130
140
Figure 2. LOS path loss consistent across bands Page 7
networks.nokia.com
1 3dB mean loss
0.8 F D C
13dB mean loss
0.6
In both cases 40dB attenuation per ray is used (73.5GHz)
0.4 0.2
Tablet/laptop device model 40dB body loss per ray
0 0
10
20
30
40
User body loss attenuation per link [dB]
Figure 3. Body loss model for tablet use case (40dB attenuation per ray at 73.5 GHz) Another important aspect for deployment at dierent frequency ranges is the penetration loss experienced by radio signals crossing various objects in the propagation environment. In the case of low frequency (<6 GHz) deployments, the main penetration loss is typically the outdoor-to-indoor propagation loss. However, in higher frequency bands, the user’s own body can also cause strong shadowing for the radio signal, in some cases fully blocking the communication link. Figure 3 shows the “tablet” body loss model that we have developed for this analysis. The radio wave propagation is modeled by multipath ray tracing with reection, diraction and scattering for each individual ray. The body loss model adds attenuation for those rays passing through the body and no loss for those rays received without interference from the user. The scenarios analyzed in this study concentrate on three models that enable comparable results:
• Simple outdoor urban case study (e.g. small subset of a city center) • Realistic outdoor urban case study (e.g. Tokyo) • Simple indoor deployment case study (e.g. oice, conference, shopping mall scenario) The next section outlines the results for LTE-Advanced, cmWave and mmWave deployments in the three scenarios.
Page 8
networks.nokia.com
METIS TC2 “Madrid” •
•
basic block 387x552m +20m extension on each side (~0.25km2) ~40% outdoor area (including park)
Deployment types: • • •
Small cells (3 ISD options, uniform deployments) Macro cells (~250m ISD) HetNet (mixtures of macro and small cells) 100m ISD (120 APs/km2)
75m ISD (250 APs/km 2)
50m ISD (450 APs/km 2)
Figure 4. Simplied outdoor deployment scenario
5G deployment recommendations The basic outdoor model is the METIS test case 2 (TC2) also referred as the “Madrid” environment [METIS deliverable D6.1] as shown in 4. The Madrid case is deployed with a macro layer having an inter-site distance (ISD) of 250 m and the small cell layer varying between 50, 75, and 100 m ISD. Figure 5 shows the capacity and the spectral eiciency of the LTE-Advanced, cmWave, and mmWave systems deployed at various carrier frequencies and various ISDs. 10
8 Thousands of
m k / z H / s / b k [ y c n e i c ffi e k r o w t e n l l u F
Gb/s/km
m k / s / 10 b T [ y t i c a p a c k r o 10- w t e n l l u F
]
]
Hundreds of
Gb/s/km
Tens of
Gb/s/km
LTE, 50m ISD cmWave, 50m ISD mmWave, 50m ISD LTE, 75m ISD cmWave, 75m ISD mmWave, 75m ISD LTE, 100m ISD cmWave, 100m ISD mmWave, 100m ISD
7 6 5
mmWave uses 4 cells/site
4 3
Efficiency of LTE and cmWave the same in DL only simulations
2 1 0
10- 10
10
10
10
Frequency [GHz]
10
10
Frequency [GHz]
Figure 5. Small cell deployment in the Madrid environment with dierent ISD Page 9
networks.nokia.com
What we can see is that LTE-Advanced is able to provide tens of Gb/s/km 2, which is expected in the early part of 2020 [Evolution towards UDN – Nokia white paper]. The next step is deployment of cmWave radio, which can carry hundreds of Gb/s/km 2 as expected from 2025. Finally, mmWave radio can provide capacity in the order of several Tb/s/km 2 which even exceeds the 1Tb/s/km2 requirement outlined for 2030. The cmWave system has a stable performance over the analyzed frequency ranges, whereas the mmWave system provides signicantly more capacity with certain deviation between the considered bands. The main reason for the increased capacity of the mmWave system is the additional carrier bandwidth and the four sectorized antennas assumed for the mmWave access points. Figure 5 shows three dierent small cell deployment cases where each provides the required capacity for a given timeframe. However, the expected deployment will be a HetNet. This will have layers of dierent cells, with a wide area LTE-Advanced or 5G system as an overlay, with 5G small cells deployed throughout the >6 GHz frequency ranges providing the required capacity when and where needed. The macro cells are deployed mainly for coverage, with an ISD of ~250m. A 100 m ISD for the small cell deployment provides full outdoor coverage. The growing need for capacity can then be provided with the deployment of dierent 5G small cells. One of the issues aecting the deployment of a HetNet is the load balancing technique for optimizing the user experience and the network capacity. In 5G systems needing to provide 100 Mbit/s, many users may be in a boundary zone, experiencing good coverage but not fully achieving the requested data rate. For load balancing, the serving cell selection procedure is based on the estimated available throughput per link, rather than simply the link quality as typically done in legacy systems. For this purpose, the user’s signal quality measurements of the potential serving cells are collected (wideband SINR measurements, e.g. RSRQ). These measurements are used to estimate the link capacity using the SINR-to-capacity mapping function. This link capacity is the throughput a single user in a cell would get. Secondly, an estimation of the potential number of other served users per cell is made based on the system type (e.g. its bandwidth) and cell size (macro vs. pico). For example, a cmWave cell serves typically four times more users than a mmWave cell, because mmWave access points have four sectors to cover the same area as a cmWave access point. Using the link capacity estimation and the typical number of users per cell, the user decides which cell will oer the higher throughput and connects to the cell. Another simple yet powerful way to increase cell edge data rates is to use dual or multi connectivity in the user devices. Table 2 shows the gain, from having load balancing with a single data connection, to having multi connectivity between cmWave and mmWave small cells. It can be seen that at 100m ISD when using load balancing, we have cell edge performance of 87 Mbit/s, which is slightly below the target in 2030. By adding dual connectivity between cmWave and mmWave, we can increase the cell edge performance by a factor of three to 286 Mbit/s.
Page 10
networks.nokia.com
Table 2. Average throughput and cell edge throughput enhancements for multi-connectivity (full buer model) Load balancing
Multi-connectivity
cmWave ISD (10 GHz band)
mmWave ISD (73GHz band)
Avg. Thp
5%-ile Thp.
Avg. Thp
5%-ile Thp.
100m
100m
1.4 Gb/s
87 Mb/s
1.5 Gb/s
286 Mb/s
75m
75m
2.1 Gb/s
210 Mb/s
2.6 Gb/s
784 Mb/s
50m
50m
3.1 Gb/s
420 Mb/s
4.1 Gb/s
1300 Mb/s
This study was based on the expected spectrum allocation in the dierent bands. An analysis was performed of how much spectrum is really needed to fulll the 5G requirements for 2030 and the minimum bandwidth required to achieve 10 Gb/s peak throughput. The results show that this is currently not possible in LTE with 3GPP Rel-12 - 5G cmWave needs 215 MHz and 5G mmWave requires 1.25 GHz. Furthermore, ~1 GHz of aggregated spectrum is required to deliver the capacity and the cell edge data rates of 100 Mbit/s for 2030. Therefore, LTE-Advanced with a 100 MHz bandwidth and cmWave with a 500 MHz bandwidth are unable to deliver the required capacity and cell edge data rates by 2030. In contrast, a 2 GHz bandwidth may be at the high end for mmWave deployment and a lower bandwidth with a higher power spectral density could be considered. However, LTE Rel. 13 standardization is working on 32 carrier aggregation and LTE-U which will enhance the data rates of LTE signicantly. Backhaul is an important issue in the deployment of small cells. Nokia has studied backhaul for 5G small cell deployments and found that mmWave deployed at, for example, 73 GHz can provide 1 Gbit/s backhaul capacity per cell in a mesh conguration with a maximum of two hops. This would mean that, in a realistic environment, only around 20% of all access points would require wired backhaul (for example macro access points) relaying the backhaul capacity to the remaining access points in the network. For a real dense urban city deployment, we have analyzed a real urban environment in Tokyo, as shown in 6. The Tokyo deployment consists of an LTE Advanced macro layer with 240 m ISD at 2 GHz representing an aggregated spectrum of 100 MHz at 2 GHz and below. Furthermore, there is a small cell layer of cmWave deployed at 10 GHz with 500MHz bandwidth and a colocated small cell layer of mmWave at 73 GHz with 2 GHz bandwidth. The initial coverage analysis showed that full coverage at the full spectrum range was not possible with the assumed antenna congurations, as the real city environment with irregular infrastructure and foliage was more diicult than the “ideal” deployment in the simplied outdoor case shown in Figure 5. Therefore, all of the small cells are deployed at 75 m ISD with the cmWave and mmWave access points co-located to provide full outdoor coverage.
Page 11
networks.nokia.com
1km
1km
Figure 6. 5G deployment in Tokyo 7 shows the coverage and capacity of a 5G HetNet deployment in Tokyo. The LTE Advanced layer provides 100% coverage, the cmWave deployed at 10 GHz provides 97% coverage and the mmWave deployed at 73 GHz provides 68% outdoor coverage. The aggregated capacity provides only 80 Mbit/s cell edge data rates for the combined three systems deployed with LTE Advanced overlay and co-located cmWave and mmWave small cells. An increase in the Base Station
User
] s p b M [ t u p h g u o r h T
s e t a n i d r o o C
Coordinates
Figure 7. Coverage and throughput maps of Tokyo 5G deployment Page 12
networks.nokia.com
macro layer spectrum to 200 MHz enabled the cell edge capacity to provide 105 Mbit/s. The additional spectrum could be deployed at 3.5 GHz, which is assumed to be released for mobile communications. Alternative solutions with more dense deployment in a few hot spot areas would also enable a 100 Mbit/s cell edge capacity with 100 MHz macro spectrum. One of the key conclusions from both the METIS “Madrid” and the Tokyo deployment was that the outdoor cells were unable to provide the neede d capacity indoors. LTE Advanced macro cell deployment at 2 GHz could provide indoor coverage but the 100 MHz combined carrier bandwidth was unable to provide the required capacity. At the same time, the frequencies above 10 GHz were unable to provide the required coverage. Therefore, it has become clear that a dedicated indoor deployment is required. Indoor coverage and capacity can be created by both cellular licensed technology and unlicensed technology. Wi-Fi evolution will continue in parallel with cellular evolution. A mixture of cellular coverage with capacity ooad via unlicensed spectrum allows seamless operation with cellular. Such a deployment can be done with integration between Wi-Fi and cellular technologies or deploying cellular technologies like LTE-U in the unlicensed spectrum. A case study of an indoor oice or conference center was analyzed as shown in Figure 8 based on 5G technologies.
Path Loss
Throughput
Figure 8. Indoor oice/conference center deployment
Page 13
networks.nokia.com
Again, similar conclusions were drawn as for the outdoor deployment case. The lower frequency band can provide coverage with a low number of access points, but it cannot provide the needed capacity, because very dense deployment will lead to high interference. The higher frequency band access points can provide the needed capacity but require a very dense deployment of basically an access point in every room or every 100 m 2 for larger indoor spaces. The 10 key 5G deployment recommendations are:
1. LTE Advanced can provide the required capacity of tens of Gb/s/km2 for 2020 and beyond. 2. ~1 GHz of aggregated spectrum is required to provide the capacity and cell edge data rates by 2030. 3. A 5G small cells deployment in the 6-30 GHz band (cmWave) with a 500 MHz carrier bandwidth can provide hundreds of Gb/s/km2 for 2025 and beyond. 4. A 5G small cells deployment in up to 100 GHz bands (mmWave) with 2 GHz carrier bandwidth can provide several Tb/s/km2 for 2030 and beyond. 5. mmWave radio can further provide backhaul to small cells in a mesh conguration with a maximum of two hops. 6. Very large antenna arrays can be used to eectively compensate for the higher path loss at higher frequency bands. 7. For both the cmWave and mmWave deployments, an inter-site distance of 75-100 m can provide full coverage and satisfy the required capacity, depending on the environment. 8. A 5G wide-area solution is needed to provide the required coverage and cell edge data rates for 2030. 9. Dedicated indoor small cell deployments are needed to satisfy indoor capacity requirements beyond 2020. 10. Multi-connectivity between LTE Advanced, cmWave and mmWave boosts cell edge performance signicantly and can lower the required density for small cell deployment.
Page 14
networks.nokia.com
Summary and conclusions Nokia studies on dense deployments in Madrid and Tokyo have shown that a 10,000- fold capacity can be provided in a dense urban environment as well as indoor dense areas. 5G will require a coverage layer that could be provided by macro cells and a coverage layer consisting of small cells providing capacity using the available spectrum range from below 1 GHz up to 100 GHz. The indoor capacity will require dedicated indoor 5G small cells. While 5G will provide a signicant boost in capacity, the deployment density of 5G outdoor small cells can be limited to ~75 m ISD and for an indoor deployment, an access point in every room is required for coverage and capacity.
Page 15
networks.nokia.com
Nokia is a registered trademark of Nokia Corporation. Other product and company names mentioned herein may be trademarks or trade names of their respective owners. Nokia Nokia Solutions and Networks Oy P.O. Box 1 FI-02022 Finland
Visiting address: Karaportti 3, ESPOO, Finland Switchboard +358 71 400 4000 Product code C401-01178-WP-201503-1-EN © Nokia Solutions and Networks 2015
networks.nokia.com
