Maximizing ARPU with RAN-aware policy management

"Aware" RANs add the intelligence to a network necessary to improve QoE and maximize ARPU.

5Network intelligence is one key to addressing service disruptions and also a means for operators to maximize their Average Revenue Per User (ARPU) to offset the increases in CAPEX and OPEX associated with network capacity expansion. ‘RAN awareness’ can provide operators with real-time network congestion information to help manage traffic flows and Quality of Service (QoS), enabling improvements in overall Quality of Experience (QoE) for all subscribers while also promoting revenue-generating tiered service models.

Smartphones and bandwidth-intensive applications, such as gaming, photo sharing, social media, and mobile TV, continue to gain popularity. A large portion of the broadband data upsurge is attributable to social media, where Twitter, YouTube, and Facebook traffic was up by between 100 and 200 percent during the second half of 2011. Trends indicate that future growth in mobile broadband traffic will continue to be driven by social media, and video in particular (Figure 1).

Figure 1: The Allot Mobile Trends Report 2011 H2 depicts the continued surge in mobile broadband traffic that was driven largely in part by video streaming in the second half of 2011.

However, mobile operator’s revenues are not increasing as fast as booming broadband demand, making it more difficult to offset their investments in network capacity expansion. In addition, they run the risk of losing subscribers due to service disruption caused by congested cells. Solutions that use Network Intelligence based on Deep Packet Inspection (DPI) and metadata extraction are surfacing as a key weapon in the mobile operator’s arsenal to combat challenges posed by exponential growth in mobile traffic. Policy management is one such application that has gained popularity as a means to achieve both cost per bit reduction and network monetization.

Wireless networks: A clear case for real-time and dynamic policies

The traditional policy management model is fairly straightforward and contains two key functions:

  • The Policy and Charging Rule Function (PCRF), which determines policy and charging rules for controlling service data flows and IP bearer resources, as well as generates Policy and Charging Control (PCC) rules

  • The Policy Enforcement Function (PCEF), which enforces PCRF policies and charging decisions by performing functions such as traffic shaping, DPI, flow marking, and Quality of Service (QoS) control; The PCEF is typically deployed at the "far edge" of the network, often collocated with GGSN in 3G networks and the Evolved Packet Core (EPC) in LTE networks

To apply this to wireless networks, it is necessary to understand the most significant pain point in delivering QoS and Quality of Experience (QoE): Radio Access Network (RAN) bandwidth is expensive and therefore limited. Given the adoption rate of smartphones and the subsequent impact on network bandwidth consumption, the traditional PCEF solution inspecting packets and data flows only at the far edge of the network largely ignores the problem in the RAN (Figure 2).

Figure 2: An example of PCRF and PCEF functions in a 3G network.

Without any real-time intelligence about cell congestion, policy rules are based on the information gathered at the Gi interface in 3G networks and the SGi interface in LTE networks. As a result, enforcement is often based on blanket policies that limit high-bandwidth applications at all times and across all cells, similar to those policies deployed in wired networks. This is sub-optimal in managing traffic because it is unable to adapt based on RAN congestion. Consequently, even subscribers on premium plans will experience lower QoS when cells are congested, leading them to question why they are paying more.

Real-time knowledge about traffic characteristics and demand in the RAN – segmented by application, user, and time-of-day – enables operators to deliver customized services that improve both their network cost profile and their ability to monetize services. For example, subscribers selecting premium rate plans, rather than “best-effort” options, can reliably be given an experience commensurate with the higher costs. Network operators need the increased control provided by RAN-aware adaptive traffic shaping and policy management to deliver more revenue-generating, tiered service models and unprecedented improvements in subscriber satisfaction aligned with Service-Level Agreements (SLAs).  Real-time “Network Intelligence” like this is one key to maximizing Average Revenue Per User (ARPU).

RAN-aware adaptive traffic shaping and policy management

RAN-aware adaptive traffic shaping is a very effective solution for mobile network congestion that uses Network Intelligence gathered from various access and core network nodes to drive real-time decisions for traffic shaping and policy enforcement. RAN-aware traffic shaping and policy management helps improve management of traffic flows during peak periods, leading to improved QoS for most subscribers and ensuring a higher QoE for those on congested cells who are paying a premium. This is achieved by adding another network function: network monitoring via the network probe.

The network probe adds cell awareness by monitoring signaling in the RAN (Figure 3). For example, it sniffs IuB/Iu-PS signaling in 3G networks or S1U in LTE networks, maps the user ID and Packet Data Protocol (PDP) sessions to the cell ID, and delivers this information to the PCRF via the diameter interface. The PCRF leverages ‘RAN awareness’ to generate PCC rules that are passed to the PCEF, which performs traffic shaping based on real-time Network Intelligence of traffic flowing through the Gi interface (SGi in LTE networks).

Figure 3: An example of RAN-aware policy management in a 3G mobile network.

Deploying real-time and dynamic policy management cost effectively and rapidly

With the strides Intel has made in data plane performance with the latest generation of server-class processors (codenamed Sandy Bridge), an entire mobile network node capacity for applications such as policy enforcement can now be serviced with just one or two Intel-based network appliances. The combination of the Intel Data Plane Development Kit (Intel DPDK) and Intel microarchitecture improvements deliver upwards of 165 MPPS of L3 forwarding throughput (64-byte packets) using the dual-socket 8-core Intel Xeon processor E5-2658, or approximately four times greater throughput than a 2009 Intel Xeon processor-based platform. In other words, a 2U network appliance can be deployed for network and protocol monitoring in multiple locations, making the PCRF “RAN aware” via standards-based protocols.

Mobile operators are now able to deploy DPI applications like RAN aware quickly and cost effectively due to the availability of relatively small form factor network appliances that are optimized for in-line or passive Network Intelligence gathering, policy enforcement, and security. Based on dual 8-core Intel Xeon E5-2600 series processors and optimized for performance with the Intel DPDK, the 2U RMS-220 network appliance server can be integrated with Trillium monitoring and Diameter stacks as a part of Radisys’ pre-integrated network probe and PCEF application-ready platform.

Critical to the sniffing/monitoring process, the monitoring solution collects data at wireline speed without disrupting the flow of traffic. The network probe provides RAN awareness via the Trillium signal monitoring software for Iub and Iu-PS or S1U interfaces. For a standalone PCEF, the RMS-220 can be deployed as a cost-effective but high-performance solution with other software products that include the Qosmos DPI and metadata extraction engine (Qosmos ixEngine) and the Radisys Trillium Diameter stack. These two deployment options provide operators with access to a wealth of information from multiple points in the network – including the RAN – as a result of data mining, profiling, and analytics, all of which are key to the creation of personalized service packages based on a deeper understanding of customers’ needs.

Figure 4: Key ingredients of the application-ready network probe

Staying "aware" of policy management

Mobile data traffic is exploding. Network operators face an on-going challenge to maximize their ARPU to offset constant increases in CAPEX and OPEX associated with investments in network capacity expansion and the adoption of advanced services. Implementing real-time policy management and adaptive traffic shaping in the mobile network via a RAN-aware PCEF can help network operators manage QoS for all users, prioritize premium customers within a congested cell, and offer tiered services to increase revenue. The availability of Intel Sandy Bridge-based network appliances, pre-integrated with Intel DPDK and enabling software, helps TEMs significantly reduce their development time and risk in delivering a cost-effective solution. As a result, the RAN-aware PCEF has emerged as an essential tool for network operators to manage mobile data traffic as it increases to incredible levels.

Chandresh Ruparel is Senior Product Line Manager at Radisys.


[1] Allot Communications. “Mobile Trends Report H2, 2011.”