How have telecommunication networks evolved over years?
The mobile network has evolved rapidly over the years. In the past a circuit switch call over a mobile network meant several network elements getting into action. Over the air transmission to antennae, base station, transmission and reception, site routers, back haul over microwave or fiber, transport network with multiple topologies, edge, pre-aggregation, and aggregation routers, core network and various operations support (OSS) and business support systems (BSS).
In 2G/GSM technology, on the radio side, you had base station, base station controller (BSC) and Mobile Switching Center (MSC), working with various TDM network- E1, STM1 interfaces. And the core network was dedicated commonly called as circuit switch core and packet core, with charging system playing a key role in real time data charging. You had a similar network for Code division multiple access (CDMA) technology with a varied set of standards but generally the same architecture.
In 3G/WCDMA technology it evolved to base station, Radio Network Controller (RNC), various TDM network, packet core, circuit switch core, and various OSS, BSS systems. Deep packet inspection started to become commonplace as the world started to move to 32 Mbps and 64 Mbps speeds with universal mobile telecommunications system (UMTS) and data starting to become more dominant.
A telecom network is a 3GPP standard-based network where the standards body decides the specifications and contours of every new technology and its evolution.
When did telecommunication networks start to change rapidly?
The entire landscape started to become interesting with the onset of 4G / Long-term evolution (LTE) when first signs of disruption and massive change started to appear. You had a network with eNodeB that replaced-older base stations, BSC, MSC and RNC. Evolved packet core replaced traditional packet core, and IP multimedia system (IMS) started to replace the traditional circuit switch core which was used for voice communication in 2G, 3G systems. IP/MPLS network started to replace the traditional TDM network or at least the newer network that was laid was 2.5 layers for transport.
LTE generally replaced 2G/ GSM networks from 1800Mhz and 900MHz bands in what is called as frequency division duplex FDD and new networks were launched in 2300 MHz, using time division duplex or commonly known as TDD. Gradually it was launched in 2100 MHz band, traditionally used for 3G and 800 MHz bands. Carrier aggregation started to bring more and more speeds.
Every new technology brought with it significant enhancements in coverage and capacity and spectral efficiency. LTE networks started to deliver up to 100 Mbps downlink and uplink speeds. The world started to consume it faster than the network could deliver.
Are telecommunication networks getting virtualized?
This article is intended though to bring forth as to how a very close-knit and standards driven multi-billion dollar industry and technology was taken over by much open, collaborative, and wider community through virtualization of every component of the network. It’s massive and unimaginable and if you look at hundreds of countrywide, undersea and global networks, impact is on billions of dollars worth of networks in a very short time.
The change started first with core network. Conversations to open packet core that was seeing a huge surge in traffic and used proprietary hardware and software started to surface. Opening up meant using x86 based hardware that could be procured from any vendor instead of proprietary purpose-built hardware provided by the same company that provided software- using open architecture, orchestration, and commercials that are not bundled. With network traffic growing exponentially, telecom operators had the incentive wherein they could leverage economies of scale rising out of common hardware and operating system virtualization layer residing in their data centers. They could build an x86 based hardware and virtualized data center and then ask vendors to bring their own virtual network functions that sit on that layer so that they could dynamically optimize-compute, memory and storage and save on money and cost. Ramping up and down resources as and when there was a surge or decline in data traffic.
What was a push for telecommunications applications that are now called Virtual Network Functions in the world of virtualization started to manifest itself in every part of the network? Evolved Packet Core which was now a virtualized packet gateway and virtualized mobility management entity could now talk to a virtual router or a virtual switch. And then many other applications such as deep packet inspection (DPI), Wireless application gateway (WAG), Broadband Network Gateway etc. that were earlier to be installed on purpose-built and efficient, in what was called as telco-grade hardware, were expected to move to open architecture based on x86 hardware.
While packet core was one big element of the network on the core or data center the other big element on core network was circuit switch core that carried voice. Media gateway and MSS were giving way to IMS to enable Voice over LTE. However, this transition from a purpose-built hardware and software to enable telco-grade reliability was increasingly being asked by the operator and the larger ecosystem to be brought in a more open and datacenter, based hardware and virtualized environment. This is now being called as Network Functions Virtualization Infrastructure.
How will telecommunication networks disrupt?
The biggest change, however, is silently happening on the Radio side. Radio takes bulk of the spend of a telecommunications network. It’s a highly 3GPP and standards driven deployment and business with wafer-thin margins. This business relies on efficiency gains achieved from economies of scale through deployments globally, having a shared production line. The purpose-built baseband and radios including cell site routers are highly efficient in packing massive compute, storage and memory in very small sizes- 0.5 U to 1 U that fits in a 19-inch rack. Radio resources traditionally have been deployed to maximize coverage and capacity and pushing the frontier. Without net neutrality, they have tried to implement quality of services to give differentiated speeds to higher ARPU or enterprise customers in various parts of the world. Or give priority to certain service offerings. Now, however, evolution seems to be happening in a very different direction. What is called as Network Slicing in 5G is basically dividing network resources into various slices wherein radio resources- basic unit radio resource block- will be divided all the way up to the core to service various slices. Each of these slices can borrow resources from other slices depending on free capacity.
What is changing in this evolution is that the same orchestrator that orchestrated VNFs or virtual network functions in the core network, IT domain or on the transport layer, will orchestrate various layers on the radio resources. So various functions eventually will evolve to become VNFs in themselves. Deployment architectures that were scaled and deployed around the world now will have huge flexibility. This may not happen in 5G, however over the coming years a purpose-built baseband or radio can be replaced by a temperature hardened, waterproof x86 hardware box deployed outdoor where VNFs, whether they are cell site routers or radio functions, will share the same hardware. This may happen sooner than expected as there are global alliances that are formed and there are deep-pocketed investors that are funding multiple start-ups.
How will telecommunication networks look in future?
Ubiquitous devices powered with a neural processing unit, latching to radio units running software on x86 based hardware sitting in enclosed temperature hardened cases designed for outdoor environments, which in turn transfer traffic on layer 3 routers which again sit on x86 based common hardware- from pre-aggregation routers to aggregation routers all the way up to a core network. All orchestrated by a common orchestrator and sitting on common virtual network functions infrastructure layer is how the future telecommunications network will look like. Standards will then be defined across industries as to how the hardware platforms will look and how routers should communicate. Industry-specific radio technology will have a lot many players that work in an open environment.
Who will succeed in the telecommunication networks in the future?
More open the network, more is the possibility that it will be dominated by a few large internet players. Imagine the same company designing a neural processing unit in an end device, building x86 hardware, writing software for routers, transport, radio and core, building orchestrator and orchestration layer and then building hypervisor for virtual network functions infrastructure layer. Branding may be same or different for end device or even a virtual assistant or any of the components of the network, but the fact that this all is created by the same company makes this company very powerful. What this does though is bring huge economies of scale to bring the cost down for several network elements and at the same time open it so that more people can contribute and benefit from it.
This is how internet started a few decades back and this is how telecommunication network will be a few decades later.