Mobile networks today are subject to increasingly onerous demand, owing to the heavy consumer use of video and data-intensive apps on smartphones and other mobile devices. And as mobile traffic volume continues to surge, attention is starting to pool on alternative forms of infrastructure, such as Wi-Fi, distributed antenna systems (DAS) and remote radio heads—all capable of creating greater density in the mobile network, increasing coverage and capacity, and improving customer experience.
Among such infrastructure, small cells are projected to enjoy the most significant growth within the short term, according to our recent report, Microwave Network Strategies and Vendor Leadership Service Provider Survey—an annual survey conducted by IHS Markit in which we interview mobile operators about their network architectures and solutions for backhaul. By 2018, mobile traffic on outdoor small cells is expected to rise to 11%, up from just 3% in 2016. Similarly, indoor small cells will handle increased traffic, from 4% to 10%, during the same period.
This is because for operators, small cells aren’t just intended to offload traffic from the macro network layer. Instead, they would be deployed in both indoor and outdoor areas where macro network coverage was sporadic and irregular, or where broadband speed was variable and inconstant. Thus, small cells would help drive traffic growth.
Critical metrics for consideration
Within our survey, we asked respondents to rate a list of service-level agreement (SLA) metrics for backhaul connectivity.
At the top of the list—rated very important by 100% of respondents—was latency, or the duration of time for information to transit from one point to another point on the network. Also meriting essential status was jitter, or the variation in time between the arrival of packets due to network congestion, timing drift, or route changes—factors that are damaging to the delivery of time-sensitive offerings, like video and rich-media services, which mobile operators hope to increasingly monetize. In turn, low jitter and low latency are critical in maintaining timing and sync required for smooth cell-to-cell or small cell-to-small cell handoff.
Jitter can be mitigated by deploying buffers, but latency can be hard to mitigate once present. And because lower latency is designed into each new generation of mobile access technology—from 2G to 3G, to LTE to LTE-A, and now to next-generation 5G—the need for backhaul networks to also support lower latency becomes even more pressing.
At present latency of approximately 20 milliseconds (ms) at best can be achieved in LTE networks. This is far from the ideal LTE target of 1 ms—which explains why latency is ranked by many as the top-rated and most desirable SLA metric.
Other noteworthy SLA metrics from survey respondents were uptime/reliability and downstream bandwidth capacity.
Here is where software-defined networking (SDN) comes into play, as traffic shifts from macro to small cells and low-power nodes, coupled with ever-more arduous demands on backhaul connectivity. Increasingly a favored approach toward achieving cost-effective, end-to-end infrastructure flexibility, SDN is, in fact, a necessary tool in order to provision sufficient bandwidth for resource-intensive applications that connect to data centers and cloud computing.
The fundamental concept in SDN is to separate the network’s control and data planes, allowing the latter to focus on its goal of moving network traffic, based on the control functionality of an intelligent centralized controller. The skillful implementation of SDN can deliver powerful benefits as well as lead to additional opportunities to uncover even greater efficiencies, potentially including the further virtualization of resources.
Challenges to overcome—but the future beckons
Nonetheless, SDN can be difficult to implement because of a number of challenges.
For one, each transport network—proprietary to begin with—is complicated by virtue of its unique setup and configuration, making management by a centralized controller difficult. Moreover, these are still early days when considering real-world deployments of SDN control across multiple network layers. Given the challenges in defining how the centralized controller intelligence behind SDN should work, more development is required before truly transparent transport networks can emerge.
The good news is that standardization might be the answer and provide the key to furthering development. This is especially true as the proprietary platforms and APIs of old are now making way for open-source SDN controllers that maximize network flexibility for future bandwidth demands.
With standards in place, the entire network infrastructure can potentially be virtualized, with a network manager seamlessly controlling via the same interface all network, computing, and switching resources as well as services. And by extending SDN to the transport layer and facilitating both multilayer and multi-domain programmability, completely new business models could be developed.
With SDN facilitated across the transport network, a more dynamic provisioning of bandwidth can be achieved—an increasingly pertinent consideration for mobile networks as small cell layers are added to the network.
True, the risk of over-provisioning backhaul capacity for small cells may exist given the likelihood of peak usage times in the network occurring alongside periods of possible significant under-utilization. Yet the SDN roadmap will also lead to more elastic and flexible transport network operations, enabling full-layer optimization of data flows between applications. This is something that simply doesn't exist today.
In my report, 5G Transport: Challenges and New Architectures, I provide more information on mobile service providers as they evaluate deploying SDN into the backhaul network. And for more information on this subject, listen to our webinar.