Overcoming Mobile Network Backhaul Challenges Posed by 5G

Posted by Mae Kowalke on Tuesday, June 28, 2016 with No comments

When it comes to 5G’s effect on the mobile backhaul network, it’s best to think in terms of revolution, not evolution. That’s the sentiment with which moderator Richard Webb (Research Director, Mobile Backhaul and Small Cells at IHS) kicked off a recent webinar—replay now available—taking on the weighty topic of how changing demands on mobile networks impacts architectural decisions about backhaul, as operators prepare for 5G.

“If we continue to build backhaul as we currently do, it will not cope for the 5G environment,” Webb stressed, kicking off a series of questions aimed at the panelists, who included Accedian VP of Strategic Marketing Scott Sumner and industry experts from Ciena and ECI Telecom.

Discussion during the webinar and Q&A covered a lot of ground, including:

  • Where content needs to be placed to accommodate 5G latency requirements
  • The economic challenges of 5G, in particular five-nines reliability
  • The role of SDN and NFV to making mobile backhaul work for 5G
  • 5G security
  • The role of fiber in 5G
  • Why moving to a flatter network architecture matters
  • How to scale C-RAN for 5G
  • Likely scenarios for development of mobile edge computing
  • Emerging backhaul technologies
  • Approaches to keeping up with bandwidth demand
  • Synchronization and end-to-end latency for 5G
Below is a summary of points made by Sumner during the discussion, focused mostly on the importance of network instrumentation as the foundation for intelligent, automated control. Watch the full webinar for a deep dive into 5G backhaul.
5G Challenges
Size, scale, and complexity are key aspects of why 5G is such a big deal for mobile network backhaul. For example, by 2020, GSMA expects early 5G networks will need to support 10 times more applications per device running concurrently, 100 times more devices, and 1,000 times more bandwidth consumption than currently. Multiplied out, this amounts to a million times more sessions than today’s backhaul networks can handle.

“Operators are realizing that this scale means some form of automation or software-defined networking is needed to manage and optimize each 5G service over this shared network,” Sumner said.

Tier 1 operators now rolling out pre-5G networks are also observing new, emerging relationships between previously uncoupled metrics. For example:

  • Small amounts of packet loss leading to a dramatic drop in throughput, impacting QoE
  • Loss burst, where small bursts of control plane signalling loss results in significant outages
  • Interactions between delay and throughput affecting QoE

As backhaul networks grow more complex, new performance dependencies emerge.

New Technologies and Solutions
Big data analytics will play a crucial role in 5G backhaul networks. This depends on uniform instrumentation across the entire network, to provide network state and QoE insight necessary for effective control and optimization.

“With virtualized instrumentation, network infrastructure itself can be used as a source for data about the performance of each application or each network slice,” Sumner noted. “That data can then be analyzed for QoE and QoS, mixed with the subscriber state, and fed into analytics and control systems.”

There are two aspects of data analytics here:
  1. Big and deep data that provides subscriber trends, leading to new service and upsell opportunities
  2. Fast data analytics, run in a RAN cloud using in-memory processing, that enables a closed feedback loop involving SDN control and service orchestration

Network-wide instrumentation is the foundation for closed-loop backhaul network control. 

The increasing virtualization of instrumentation—in particular, distributed packet brokering technology that pulls insights from every layer and location—makes it possible for centralized analyzer and SDN control systems to pull full network state information from all corners of the network, and take appropriate action.
The Importance of Quality of Experience (QoE)
For operators to succeed with 5G, they must have a way to view and control the full lifecycle of each service. That means assessing network readiness pre-rollout, validating performance at turn-up, and continuous QoE assurance post-launch.

That last step is especially noteworthy because it means the operator effectively closes the feedback loop, validating predicted performance trends and tracking per-user QoE.

“Increasingly, operators are less interested in making sure the network mechanically is working perfectly for every possible dimension and metric, and more interested in the user experience, which becomes the ultimate judgement of whether the network was deployed properly or not,” Sumner concluded.

For example, SK Telecom uses SDN control and intelligent backhaul pathing to measure the network state across all backhaul links and move traffic around as needed to maintain optimal QoE.

“By measuring each link’s performance, SK Telecom can determine which particular link is appropriate for which application,” Sumner explained. “For example, one link might be very slow but have very low latency, and therefore quite suitable for sending control plane signaling. Whereas, another link might be fairly high delay but has a lot of bandwidth available, which might work very well for video streaming. It’s a matter of matching up the characteristics of each application or service with what’s available in the network so you can use every ounce of capacity as efficiently as possible.”

Operators can use the same logic to determine how they should scale up virtual network functions, where to position VNFs in the network, and assure the availability of resources to support good QoE.


For network backhaul design, what do we need to stop doing and start doing to prepare for 5G?
In Asia, some operators are intent on rolling out pre-standard 5G networks in the next couple of years. The example they’re setting is a good place to look for guidance on backhaul design. For example, two ways they’re preparing are:
  1. Stop treating everything as a single class of service or use over-provisioning to keep up with bandwidth demand. 
  2. Pay attention to the effect network traffic microbursts are having on impairing flows and QoE; respond by instrumenting the network so it can be monitored with sub-second precision in real time.
“Network-wide QoS and QoE visibility is the foundation operators will use to build SDN control with analytics on top,” Sumner concluded. How can an operator get started with establishing a fully-instrumented network that interface with SDN control, big data, and analytics? How long will this take to deploy?
Using traditional probes, this would be a couple year project. But, with the virtualization of instrumentation, that timeline collapses dramatically.

“We’ve seen networks now in the quarter million endpoints range being instrumented in 8-12 weeks,” Sumner said. “This is possible using highly virtualized instrumentation, running on commercial off-the-shelf servers, and using the existing network elements as endpoints. This is becoming a much simpler endeavor and much faster for the operator because they can use their existing stack to get things up and running quickly.”


Video Playlist: Network Intelligence for 5G
Video: FlowBROKER: Network-Wide Analytics, Delivered
White Paper: 5G Mobile QoE
Video: Demo: Accedian FlowBROKER for QoE Troubleshooting
White Paper: Decision Factors in Service Providers’ vCPE Migration
Infographic: 5G QoE: Under Siege?

Categories: , , , ,