By Gautam Billa
Mobile broadband is pivotal for Indian telecom sector growth in the next 3-5 years. India is still only 26% internet penetration (global average is 44%). It is estimated that India will have in excess of 600 million mobile internet users by 2020. Network coverage and handset affordability are two key gating factors in this journey. While 5G as a technology is expected to revolutionize the telecommunication domain and beyond yet again, it will be techniques such as Network slicing / resource virtualization and densification that will play a fundamental role in building these next generation networks in India.
One of 5G’s biggest and most heralded promises is that there will be greater bandwidth for all – as much as 10Gbps peak rate per user, which is truly enormous. 5G also promises as much as 99.999 percent network availability, theoretical 100% coverage, and a dramatic decrease in power requirements, up to as much as 90 percent reduction compared to today’s technologies. Users may claim to need all the bandwidth they can get, but that doesn’t necessarily correlate with their actual requirements.
That being said, there is also something that is fundamentally unique about 5G that was never seen in any of the earlier generations. The core thought process of 5G is built around “services” and not around technology. Below are three main services classifications under 5G:
Enhanced Mobile Broadband (eMBB)
- High speed browsing and downloads
- 4k / 8k video streaming
Ultra-reliable & Low Latency Communications (UR-LLC)
- Self-Driving Cars
- Manufacturing Process Automation
Machine to Machine Communications (mMTC)
- Smart Cities
- Automated Homes
From a network design perspective eMBB is all about high speed data, throughput, spectral efficiency, and peak data rate. UR-LLC can do with slightly lower data speeds but these need to have 100% coverage on-the-move and lowest possible latencies. mMTC has another different set of requirements; it needs hyper-scale in terms of number of devices (typically sensors in IoT terms) and energy efficiency – the billions of sensors cannot be “always on”.
The typical rules that govern traditional approach of network design tell us that the parameters that we just described above are often contradictory and require trade-offs. If we tune the network for best performance on one aspect, we need to forego performance on some other aspect and vice versa. We can’t have everything at the same time.
The need for Network Slicing
5G will need all these at the same time and is cost prohibitive to have different physical networks for each template of requirements. This is where network slicing or resource virtualization enters the fray. Network slicing allows a service provider to create multiple virtual network slices over the same physical infrastructure. Slice-1 can be optimized for high reliability and lowest latency to support an autonomous vehicle network. Slice-2 can be designed for sensors installed in a farm to periodically record temperature and humidity readings. Network slice extends right from the wireless domain into the core up to a data center. It may have its own virtual EPC, its own virtual firewall, its own virtual transport network and its own independent address space.
This opens up the possibility of viable, sustainable multi-tenant networks. Using a slice-by-slice approach, the Service Provider can allocate each tenant their own network slice, completely isolated from other slices and other tenants. Tenants will also be free to better analyze data, as well as define their own service levels, quality of service, and charging metrics, providing wholly
Navigating the challenges in India’s telecoms landscape
Chief among challenges are two that afflict many change or migration initiatives. The first is that migrating away from legacy structures will always be fraught with risk if not undertaken with the utmost care. The second point is that with the adoption of any new technology comes the need to acquire new skills; from software-defined networking to customization and virtualization. As universal as these challenges are, network vendors are available to help MNOs train their staff to appropriate levels.
In India in particular, some of these challenges may be felt more keenly. With so much hope pinned on the revenue-generating possibilities of mobile internet growth, the sector has a clear need to invest in the future. But against a backdrop of an ongoing price-war to attract net new customers, in a market where coverage isn’t always guaranteed and handset cost can be a barrier, many of India’s network operators could find themselves caught in a catch-22 scenario; you need to invest so you can benefit from the uptake of new data-led services, but such investments may seem just out of reach.
The adoption of network slicing will need to be handled in phases, considering the existing legacy infrastructure and any elements of the extended physical network that might fall outside of an operator’s immediate and direct control. This will require careful consideration to ensure extending the network slicing out from owned assets to non-owned assets doesn’t result in any breaks in the slice. At the back-end, OSS/BSS systems may need to be restructured, or even replaced, to ensure appropriate billing and operational support.
In a world that has fallen in love with cloud computing, there are still people objecting with, “but distributed computing models are as old as the hills.” The potential for using virtualization within the sphere of network slicing, could also be dismissed as nothing especially new – especially to those familiar with deploying VPNs across LAN and WAN environments. But with the promise of so much, 5G will require a new way of thinking when it comes to virtualization – hence network slicing has the potential to unlock new services, set free existing ones, and support those that have yet to be envisioned.
(The author, Gautam Billa, is Director Engineering at Ciena India)