The Main Transformative Aspects of 5G
In this installment of the IEEE 5G Transmissions: Podcasts with the experts, we talk about the main transformative aspects of 5G with Dr. David Soldani. Dr. Soldani is an IEEE Senior Member and Associate Editor-in-Chief of IEEE Network Magazine. He is the head of 5G technology, end-to-end and global for Nokia Germany and he is an industry professor at the University of Technology in Sydney, Australia.
5G defined as the International Mobile Telecommunications for 2020 and beyond (5G) will expand and support diverse usage scenarios and applications with respect to current mobile network generations, purposed primarily for voice, mobile internet and video experience.
The agreed scenarios for 5G include:
1) “Enhanced mobile broadband (eMBB)” addressing human-centric use cases for access to multimedia content, services and data;
2) “Ultra-reliable-low-latency communications (URLLC)” with strict requirements, especially in terms of latency and reliability; and
3) “Massive machine type communications (mMTC)” for a very large number of connected devices and typically transmitting a relatively low volume of non-delay-sensitive information.
5G technologies will efficiently enable new secure, dependable, ultra-reliable and delay-critical services to everyone and everything, such as cognitive objects and cyber-physical systems.
To realize this vision, 5G capabilities will include: A new flexible and efficient wireless interface, access schemes and other enabling wireless and network technologies, as well as a new plastic network architecture, supporting multi-tenant and new role models. With network slicing, different end-to-end logical networks with isolated properties are provided and operated independently. These enable operators to support different use cases, with devices able to connect to multiple slices simultaneously, and monetize network slice instances as a service.
Two architectures options are possible:
- The radio access will be based on New Radio (NR) in a Stand-Alone (SA) configuration of 5G systems – Devices, Next Generation (NG) Radio Access Network (NR), NG Core (or 5GC);
- or Non-Standalone (NSA) NR in an Evolved Packet System (EPS), which is a dual connectivity 5G System deployment with an Evolved-UTRA, as the anchor Radio Access Technology (RAT), and NR as the secondary RAT, in a non-standalone configuration in Evolved Packet System.
5G is expected to increase spectrum efficiency and support contiguous, non-contiguous and much broader channel bandwidths than available to current mobile systems. 5G radio will be the most flexible way to benefit from all available spectrum options from 400 MHz to 90 GHz, including licensed, shared access and unlicensed, FDD and TDD modes, including supplementary uplink (SUL), narrowband and wideband allocations.
Millimeter wave spectrum above 20 GHz can provide bandwidth up to 1-2 GHz, which offers data rates up to 20 Gb/s and extreme mobile broadband capacity. The high bands are mostly suitable for local usage, such as mass events, indoor and outdoor hotspots and for fixed wireless access (FWA).
Spectrum at 3.5 GHz, 4.5 GHz and 4.9 GHz will be used for 5G coverage and capacity in urban areas reusing the existing sites. At those frequencies, the bandwidth can be up to 100 MHz per operator, and even up to 200 MHz with the re-farming of some of the existing bands. In fact, 5G coverage at 3.5 GHz, when using massive MIMO beamforming, can be similar to LTE 1800 or 2100.
Low bands, below 1 GHz, are needed for wide area rural coverage, for ultra-high reliability and for deep indoor penetration. Extensive coverage is important for new use cases such as IoT and critical communication. The low band could be 700 MHz, which is available in many countries, or 900 MHz, today mostly used by 2G and 3G systems. In the USA, another option for 5G low band is 600 MHz.
The new 5G radio is for all spectrum options. To this end, 5G supports a flexible numerology, which consists of different sub carrier spacing, nominal cyclic prefix, and transmission time intervals, or scheduling intervals, depending on bandwidth and latency requirements. Sub-carrier spacings of 15 kHz to 120 kHz, and the corresponding cyclic prefix of 4.7 to 0.59 µs and scheduling interval of 1ms to 0.125ms, are defined for different carrier components, which may vary from 5 MHz to 100MHz, below 6GHz, and from 50 to 400MHz, above 6GHz.
For optimal radio performance, the higher the carrier frequency, the higher the allowed carrier component and subcarrier spacing, the lower the corresponding cyclic prefix and scheduling period. If we consider a typical 5G deployment at the 3.5 GHz band, the bandwidth could be 40-100 MHz, the subcarrier spacing 30-60 kHz and minimum scheduling period 0.125 ms. The corresponding numbers in LTE are 20 MHz bandwidth, 15 kHz subcarrier spacing and 1 ms scheduling period. In the narrowband cases where low latency is required, the so called ‘mini-slot’ can be used in 5G, where the transmission time may be reduced to 2, 4 or 7 OFDM symbols. It is also possible to combine multiple slots together.
Massive MIMO (mMIMO) will be an integral part of 5G from day one, including common and control channels with beam forming and tracking. With mMIMO the number of transmitting antenna elements is much higher than the number of MIMO streams (or layers). In practice, mMIMO means that the number of controllable antenna elements is more than eight.
5G radio will support 8 Layers Single User (SU)-MIMO or 16 Layers Multi User (MU)-MIMO in the downlink, and 4 Layers SU-MIMO in the uplink, with the possibility of dynamic switching in both directions. Multiuser MIMO means that parallel MIMO data streams (or layers) are transmitted to different users at the same time-frequency resources. A typical example of 16 Layers configuration is 2layer/UE×8UE MU-MIMO.
Beamforming offers the advantages that the same resources can be reused for multiple users in a cell: It allows Space Division Multiple Access (SDMA), maximizing the number of supported users within that sector: It minimizes interference and increases cell capacity. As a result, Massive MIMO and active antenna technologies (AAT) give higher spectral efficiency and provide much more capacity on existing base station sites. The technology can also enhance link performance and increase the coverage area.
5G radio design is fully optimized for massive MIMO using three basic techniques for forming and steering beams:
- Digital beamforming, where each antenna element has a transceiver unit with the adaptive Tx/Rx weights in the baseband, enabling frequency selective beamforming. Digital beamforming boosts capacity and flexibility and it is mostly suited to bands below 6 GHz.
- Analog Beamforming implements only one transceiver unit and one RF beam per polarization. Adaptive Tx/Rx weighting on the RF is used to form a beam. This is best suited for coverage at higher mmWave bands and offers low cost and complexity.
- Hybrid beamforming is a combination of analog and digital beamforming. When some beamforming is in the analog domain, the number of transceivers is typically much lower than the number of physical antennas, which can simplify implementation, particularly at high frequency bands. This technique is suited to bands above 6 GHz.
The radio interface protocol stack is also optimized for cloud and distributed computing, with a flexible front-haul split, between high-layer protocols (PDCP and RLC), and low-layer (layer 1) peer entities, using the Enhanced Common Public Radio Interface (eCPRI). This will allow to relax the transmission capacity and the utilization of Ethernet. For example, assuming 100 MHz band, a 3-sector site with 64TX/RX mMIMO and 16 layers, the interface between the radio unit (RU) and edge cloud (radio access unit, RAU) would require: 1Tb/s with no split, using a CPRI interface; 150 Gb/s with low layer split and enhanced CPRI (eCPRI) interface; and 1-10 Gb/s with high layer split. The latency with low layer split is expected to be below 0.1 ms, as with CPRI; and 5 ms with high layer split, which is still satisfactory for ultra-reliable low latency communications (URLLC).
Compared to LTE1800 with 2x2 MIMO, as baseline, 5G gives 10-20x additional capacity, being 2-4 times more spectrally efficient. For example, LTE with 20MHz band achieves 40 Mb/s cell throughput, which corresponds to a spectrum efficiency of 2b/s/Hz; 5G with mMIMO beamforming at 3.5 GHz and 100MHz band reaches 400-800 Mb/s cell throughput, corresponding to a spectrum efficiency of 4-8 b/s/Hz.
The radio interface latency with 5G is also dramatically reduced compared to previous releases. The target is 1 ms in idle, and connected mode with and without uplink resources allocated. 5G solutions to low latency are: Shorter transmission time interval, connected inactive state and contention based uplink.
5G is not only about radio!
Additional benefits will come with the introduction of the 5G core, which supports many new enabling network technologies. For example, the 5GC is characterized by a layered and service oriented architecture, with control plane and user plane split and shared data layer, for subscription, state and policy information. It also supports: User plane session continuity, while the terminal moves across different access points; interworking with untrusted access; a comprehensive policy framework for access traffic steering, switching and splitting; and wireless-wireline convergence.
Other fundamental 5G enabling technologies, end-to-end, are: Flow-based QoS, with a much higher level of granularity than LTE, which is currently limited to the bearer service concept; multi-connectivity, where the 5G device can be connected simultaneously to 5G, LTE and WiFi, offering a higher user data rate and a more reliable connection; terminal assisted Network Slicing, and end-to-end (E2E) network management and orchestration, with in-built support for cloud implementation and edge computing.
5G Network Slicing comes along with new information and role models, and slice management functions, responsible for the management and orchestration of network slice instances (NSI). An NSI consists of one or more network slice subnet instances (NSSI) of physical network functions (PNF) and/or virtualized network functions (VNF).
Within this framework, three main role models are defined, namely:
- The Communication Service Customer (consumer, enterprise, vertical, CSP, etc.), who may use communication services from a Communication Service Provider (CSP);
- the Communication Service Provider builds its offering on top of network services, from the Network Operator, and virtual infrastructure services, from the Virtual Infrastructure and Data Center Service Providers; and
- the physical and virtual network functions composing the network slice instances, end to end, may be provided by Network Equipment Vendors (including VNF), Network Function Virtualization Infrastructure Suppliers, and Hardware Suppliers.
It is of course intended that an organization may play one or several roles.
The new CSP offering, enabled by 5G Slicing, is Network Slice as a Service (NSaaS). Like cloud computing SaaS, IaaS and PaaS models, the Communication Service Customer, i.e. the tenant, may compose, order and pay one or more network slice instances based on its utilization; service level agreements (SLA), e.g. in terms of latency, throughput, and reliability; and value-added services (VAS).
In practice, 5G will support three basic business models for network slicing, depending on the tenant’s degree of slice control, which may go from monitoring only the KPIs related to the signed SLAs, changing the configuration of the deployed slice instances, to chaining own physical/virtual network functions.
The partitioning model may be combined with the layering model to provide joint horizontal and vertical offerings. In slice partitioning, the orchestration of resources and capabilities, from an E2E service requirement perspective, must be horizontally federated (cooperation/collaboration), and vertically coordinated (hierarchy) through policies and standardized Interfaces/APIs.
The E2E 5G systems will consist of six domains, from terminal (UE) to the data network (DN) and service applications:
- Terminals: supporting Network Slice Selection Assistance Information, to request specific slice instances, based on the communication services in use.
- Access: eLTE/NR Radio Units with Ethernet front-haul (eCPRI) or ethernet mid-haul for low latency and latency insensitive services, respectively.
- Aggregation: Radio Clouds with their own Software Defined Network Controllers (SDN-C) and Virtual Infrastructure Managers (VIM).
- Mobile core: Core Cloud with own SDN-C and VIM interconnected to the Radio Clouds by IP routers and WAN SDN-C.
- Network Slice Management and Orchestration: An E2E Service Orchestrator for the embedding of Network Service Descriptors (Network Connectivity topology and VNF forwarding graphs), on top of a Self-Organizing Network (SON) and VNF Manager functions.
- Data Layer and Application Enablement: g. IoT and Customer Experience Management (CEM) platforms for running applications on top of the different network slices for public safety, digital health, mobility, industry automation, smart cities, etc.
Artificial Intelligence, in terms of descriptive, predictive and prescriptive analytics, will find application in three main areas:
- SON: Key capabilities / algorithms / architecture attributes within the different domains (RAN, Core, Transport etc.) to enable the right flexibility and tradeoffs for operators to efficiently exploit slicing;
- Data and application layers, i.e. big data analytics (structured data analytics, text analytics, web analytics, multimedia analytics, network analytics, mobile analytics), and
- Data-layer platforms for IoT and CEM.
To realize the 5G vision I have just presented: spectrum must be made available first, global standards, next, and regulations must follow. It is also necessary a massive investment from industry, especially from Connectivity Service Providers (Operators).