Braving the Signalling Storm

Fourth generation network technologies are well and truly here, and several operators have already experienced outages in their LTE networks due to signalling ‘storms’ in various parts of the network. The explosion in traffic and bandwith demands is well acknowledged but signalling is another issue operators will have to contend with. LTE networks will have 100 per cent smartphone penetration from day one and smartphones generate significantly more signalling traffic than feature phones or data cards, both on the access network and the core network.

Aside from “chatty apps” on the device, such as Facebook, Skype or WhatsApp, operators can expect the network core to see an uptick in signalling traffic as they seek to introduce richer services such as VoLTE and RCS, which send lots of policy requests in the signalling channel. This will be exacerbated by the move to real-time, dynamic charging and shared usage plans, which will increase the traffic between operators’ policy engines (PCRF) and various other elements in the back office.

Tasked with helping operators address this issue is Mallik Tatipamula, Vice President for Service Provider Solutions at F5 Networks, who joined the firm at the start of 2013. Tatipamula’s resumé reads like a who’s who of Silicon Valley, with Motorola and Cisco among his previous employers. He most recently held the position of head of research at Ericsson Silicon Valley.

Along with working alongside most of the world’s tier one operators over the years, Tatipamula believes his background has given him a good mix of understanding of both the telecom and IT/networking worlds and allowed him to spot the convergence of telecoms and IT early on. “My intention was to learn and expand into these new technologies and help shape next generation networks,” he says. “Right now, service providers are transitioning from connectivity to experience providers, while transforming their services and business models

“But in the past to the present scenario, service providers consolidated their transport infrastructure towards all IP mainly driven by fixed mobile convergence. So, In the process, fixed mobile converge became focused on transport convergence, so every operator I know has consolidated their infrastructure , but they have only looked at the control aspect in part and the services aspect not at all. There is no full convergence here,” he says.

In terms of discussion over the control plane, the industry still has IMS, but this is mostly implemented on the fixed side, although there are ongoing discussions on using voice over IMS for LTE and deployments may even start taking place this year, using VoLTE on IMS.

But largely, the control aspects of services are still to be converged, and between the services and the control layer the industry is seeing internet services like Google and the OTT providers arrive in the middle, offering the same services and applications but on an access agnostic medium. For example, users can get their Gmail on a laptop, desktop, mobile device or tablet, over fixed or mobile access. OTT players are able to provide apps to any device.

“So next I see IT and telecom convergence getting ready for content delivery in the cloud. We’re moving to a future mode of operation where we see control and services consolidated into cloud and a common IP transport also cloudified.

“Cloud is really nothing but a combination of storage, processing, and networking. So the moment you add caching (storage or memory elements) and transcoding (a processing element) on top of the RAN or EPC (Evolved Packet Core), it becomes cloud. “The future now is fully converged application delivery networking, with the convergence at the transport , control and services layers for different types of access mechanics,” Tatipamula says.

According to Tatipamula, convergence today is mainly driven by application delivery, but in the past it was driven by technology. Now the focus is more on business models and service models, which is where application delivery networking comes into the picture. In Tatipamula’s words, the industry is “moving from technology centric and network centric convergence of the past to application delivery centric convergence in the future.” The question many operators are asking now is; how can I deliver applications at lowest TCO while still increasing ARPU? The answer, according to F5, is by improving the quality of experience to end users and enabling new applications, allowing service providers to move into new verticals like automotive.

It’s a business formula F5 refers to as the ‘three Es’: economics; experience; and enablement, all of which need to be optimised, secured and monetised.

“When it comes to experience and enablement, we can do application optimisation, content optimisation, and personalisation through user, application and network awareness, and as a part of our strategy, we will soon be announcing mobile device and application management, so operators can differentiate enterprise apps on different devices,” Tatipamula says.

When it comes to economics, application optimisation is F5’s bread and butter, with traffic optimisation on the data plane carried out using loadbalancers. The second thing the company does is “signalling optimisation” to help manage the exponential growth of smartphone signalling using tools aquired with Traffix Networks. As operators are discovering, it’s important to optimise data plane traffic, signalling traffic and application plane traffic.

As operators move to LTE, legacy signalling protocol SS7 is being replaced by purely IP-based signalling interfaces, such as Diameter. Diameter is an authentication, authorisation and accounting protocol for networks, which also supports mobile management.

Diameter is largely a new protocol for GSM operators, although it will be familiar in part to CDMA operators, which already have some experience of its predecessor, RADIUS. However, all operators need to invest in the right sort of Diameter equipment in network. As a standard, Diameter is relatively stable but can quickly become complicated once an operator moves on from just using it for user authentication to using it to manage roaming users in all possible scenarios.

Diameter is the protocol of choice for doing everything and anything in LTE. This includes connecting data plane components like the PDN gateway to control plane components like PCRF to application plane components such as IMS. Diameter then connects those things to backend infrastructure such as B/OSS and online charging systems. But then, contrary to the other signalling challenges that operators are facing, Diameter is associated with revenue generation since it is often coupled with policies and charging. By default, Diameter has become a much bigger priority for operators compared with legacy signalling, which primarily drove network functionality (call setup) and operators only invested in handling signalling traffic. Mallik Tatipamula, Vice President for Service Provider Solutions at F5 Networks But now operators are deploying Diameter routers to operate a more efficient network, with Diameter edge agents for roaming, Diameter Gateway for interworking of 3G and LTE networks,and load balancing to make sure critical components do not fail. It is clear that Diameter is not a simple protocol to handle and operators are presently grappling with different deployment choices. The message from F5 is that operators need to optimise their control plane and scale with intelligence.

Personalised, media-rich applications and mobile devices are putting an enormous strain on the signalling network, which manages subscriber information, network status, and session management. So service providers need a scalable, reliable control plane to satisfy customers across all signalling interfaces such as Diameter, RADIUS, DNS, and SIP.

Although the focus is currently on Diameter, SIP is also expected to create challenges at the edge of the network when LTE-specific services, including VoLTE and RCS, enter the mass market. VoLTE signalling flow is significantly more complex compared with legacy voice and estimates from Informa Telecoms & Media indicate that VoLTE SIP requires at least double the signalling messages that a “vanilla” SIP application requires. Additional LTE services (including SMS) are expected to exaggerate the number of signalling messages even further.

“Complex signalling is a key issue in modern networks because the tear down and set up of sessions happens so much,” says Tatipamula. “Signalling is a killer in mobile networks. But with multiple apps running on smartphones and smartphone adoption growing, signalling pain will only continue.” In a sense, the volume of traffic can always be solved in terms of the data plane by putting the traffic through traditional optimisation. The solution for signalling needs to be more creative.