According to industry analysts Juniper, 60% of mobile data traffic will be carried, not by conventional mobile networks, but by WiFi . The network of tomorrow will be a mix of LTE, legacy technologies such as 3G, and a diverse set of small cells that leverage WiFi and related technologies.

This means that MNOs must increasingly consider how they will route traffic in their network. Until now, users have often unilaterally selected WiFi access to obtain faster download speeds, avoid costly roaming charges, or to bypass their plan limits. This manual selection has the result that they typically leave their MNO’s network for a competing WiFi solution where one is available.

By deploying a greater mix of WiFi, small cells and macro cells, in other words, moving to heterogeneous networks, MNOs will increasingly include alternative access technologies as part of their offer. This will help them retain customers and fully leverage their investments.

But, to ensure that users leverage these, MNOs need to help them in network selection. They need to be able to seamlessly connect users to the best available network and to do so selectively, based on the demands of different applications and traffic types.

This is where the ANDSF comes in. It’s early days for this new(ish) entity, but it is likely to assume a central role in future networks. Juniper’s research implies a significantly increased role for the ANDSF, as something has to manage the selection of the most appropriate access technology and this is the most likely solution.

It’s not going to be sufficient, however, to simply deploy centralised ANDSF solutions. MNOs will need to extend ANDSF and TDF functionality into mobile devices, to ensure that traffic management and access network selection decisions are made according to local conditions and demands.

The network is evolving and migration to LTE will lead to many changes. But headlines about the growth in mobile data traffic don’t tell the whole story. MNOs need to consider the likely mix of data traffic on their network as well as the kinds of applications that drive it. This level of detail will help them determine the appropriate mix of access technologies in their (heterogeneous) networks. MNOs will need to explore how a distributed ANDSF architecture can help them achieve optimal traffic routing. Device-based ANDSF solutions will complement those in the network, helping deliver the best customer experience and ensure that users stay connected to the MNO network.

Traditionally, network performance monitoring is provided by network probes or statistical analysis of performance from network equipment. These elements are deployed in the network and are able to analyse traffic streams, detect anomalies and generate the data that can be used by analytic applications to support service assurance and diagnostic applications.

The emphasis here is on the network view: despite the capabilities of such equipment these solutions are all network based. It is network centric, not user centric. This means that true end-to-end performance management is beyond their capabilities, as they only obtain a view of performance from the perspective of the network. They do not and cannot capture or gain an appreciation of the end user’s view or what actually happens locally in the applications that are utilised. This lack of user and application perspective and awareness is likely to become an increasingly important issue.

What’s more, traditional performance monitoring solutions are tied to a specific access network, whereas, today, a user’s experience is based on their use of multiple access technologies (e.g. 3G, LTE, WiFi, WiMax). The ability to monitor user and application views of performance across these technologies requires that the solution becomes more ‘device view’ centric as opposed to network centric.

Growing consumer and enterprise demand for smart devices, coupled with the continuing performance growth of such devices, means mobile users are increasingly using applications that have very specific requirements in terms of the carrier network. Maintaining Quality of Experience for users is increasingly difficult. That’s why we need to include the smart device within performance monitoring solutions.

Incorporating user devices in an end-to-end solution provides a user’s and even individual application’s view of actual performance. In fact, while it’s necessary to monitor conditions experience by users on their devices, this is still not sufficient to deliver the kind of end-to-end experience that is required.

We need solutions that can also respond actively to instructions from network-based solutions so that in addition to reporting on local conditions, they can proactively help manage traffic to optimise sessions. In reality, this means parsing, filtering and tagging data so that reports can be generated and action taken by other equipment in the chain. In turn, the smart devices can respond to instructions that are provided from the network, through policies enabled on the smart connected device.

Performance monitoring begins in the device. If we want to deliver a truly optimised end-to-end experience, it’s an inescapable conclusion that we have to incorporate devices into our plans. This means deploying solutions like the GoS Agent, which takes performance-monitoring functionality and enables it through policy control on user devices.

With such a solution, users can benefit from a co-ordinated, optimised policy and experience management infrastructure, and operators benefit from end-to-end visibility of the issues users really encounter in the field. This, coupled with the ability to actively correct and optimise conditions for users is what will, finally, deliver the end-to-end experience that users demand.

The primary reason given for this is that it can lead to significant efficiencies in spectrum utilisation. Although in the long term, 4G gives many benefits in terms of achievable speed, there is still a need for network diversity and small cells to complement macro sites. By using a mix of small cells that do not use valuable 3G and 4G spectrum, operators can use the spectrum that they own more efficiently.

To benefit from these efficiencies, MNOs need to develop a WiFi offload strategy. But it’s not sufficient to install small cells and simply obtain additional coverage, relying on user selection. They need to be able to make the process of switching between WiFi and 3G / 4G networks seamless, so that customers are given dynamic and automated routing options.

This is where the ANDSF comes in. It’s a new element that helps co-ordinate the selection of available access networks in line with the policy of the network operator. By deploying ANDSF solutions, MNOs can enable the automatic selection of adjacent WiFi networks for either all traffic from an individual user or even on a dynamic basis, according to the needs of each application.

Such solutions can dramatically simplify choice for users, enabling automatic registration and seamless handover, depending on configuration by the MNO. The ANDSF is essential to implementing an effective WiFi offload strategy.

However, to ensure optimum performance, the ANDSF must be able to support different policies for different users. One user might have a different price plan and access rights from another. In this case, ANDSF functionality needs to be available at the level of individual user devices.
GoS Networks’ ANDSF solution adopts the innovative step of decomposing the functionality between the GoS Agent and GoS Manager. The GoS Agent acts as distributed ANDSF client that can be deployed on mobile devices (UE), while the GoS Manager provides co-ordination, policy administration, distribution and backend database functionality securely in the network, managing all active distributed ANDSF clients via the S14 reference point.

By deploying distributed ANDSF clients under the co-ordination of a central server in combination with GoS Networks’ distributed PCEF functionality, MNOs can establish a range of policies to meet the dynamic needs of the network and users, obtain visibility and control of local conditions, and increase user satisfaction and customer experience.

If WiFi is to really make a strategic contribution to MNOs’ businesses, they need to consider how this can be achieved. Only by complementing their small cell deployments with the means to integrate fully into their policy framework will they be able to maximise returns from their investments. Deploying a distributed ANDSF client is a necessary step to reaching this goal.

Here are some of the “hot topics” that we have uncovered:

Analytics

The need for a “feedback loop” involving real or near-real time analytics that accompanies device-level policy control is likely in the future if MNOs want to apply appropriate policy decisions. We agree, along with many others, that the device is one of the best places to create both an end-to-end and customer-eye-view of the network, specifically of its performance at the application level. We also agree that it is essential to be able to make these performance measurements independent of the Radio Access Network. For example, a rule that forces a device to always choose WiFi over 3G (when available) may not be appropriate until the performance of the WiFi network can be established in real-time and then applied to the policy decision.

WiFi Offload and Distributed ANDSF

Taking bulk data away from the network benefits both the operator and the user in that it reduces strain on the network and it also reduces the use of the subscriber data bundle. Essentially, MNOs need a solution that delivers excellent network connection from those available and that optimises user experience.

WiFi hotspots and ANDSF were both big points of discussion. The key drivers behind these two elements are the ability to reduce the cost of service delivery, and the availability of high speed WiFi technology to improve user experience in congested inner city cells.

Additionally, through device-level policy control, decisions can be made at an application, deciding which applications should be used over which access technology. For example, apps running on the device can be managed by the policy software before they attempt to connect to the network and before the signalling process begins.

Watch out for more insight soon in part 2.

From the obvious controls to stop little Jane or Johnny from seeing the ‘wrong’ sort of websites to controlling and limiting access to social media and messaging services at school time or late at night.

A large number of these are connected to network based services for filtering traffic from the device using network DPI and policy enforcement points to detect traffic, classify it and determine its appropriate use and apply the parental policies to the traffic.

The down side of this approach is that it works really well while Jane and Johnny are on the mobile network, but as soon as they log onto a public WiFi hotspot or if they are abroad on a roaming network the controls may no longer work.

One of the advantages of putting policy control onto the smart connected device is to create enforcement that is network independent, so the service provider (in this case a parental controls solution from the cloud) can deploy policies that will work on the device at an individual level irrespective of which service providers network is giving them internet access.

To this end, policy control solutions on the smart connected device can ensure that Jane and Johnny remain safe and secure while they are browsing the internet in their hands – no matter where they are and whose network they are attached to. This is a great example of OTT policy control between client and server, or from the Cloud, as we now describe everything in data centres somewhere.