For over a decade, Operators have been struggling to understand the applicability of Policy and charging rules function(PCRF) in the network. The PCRF was initially envisaged to be "Brain of the network" - a centralized policy decision point for the entire network. Instead, with quota management inclusions, It seemed more like a cheaper option for OCS. Furthermore, 4G LTE and IMS with a predetermined quality of service characteristics used PCRF as translator function to map SDP to PCC rule information.
3GPP started off with defining PCRF to be an optional component of the core network. Access implemented quality of service policies on best effort basis, limiting PCRF policies primarily to the core. QoS control at the core seemed ineffective and reduced the role of PCRF to fair use and throttling.
It seems with 5G, 3GPP is attempting to fix the role of the policy control function in the network. For starters, the 5G Policy Control function (PCF) role is not limited to session management alone. The operator can use PCF function to implement session, access and also UE policies via a single policy function. With policy control scope extending to all aspects of the network, Operator can manage network resource to deliver the best subscriber experience. The same engine can also be extended to influence access selection, trigger session offload, and enforce mobility restrictions.
One of the key change in PCF decision making has been its ability to include analytics (from both core and access nodes). This input allows PCF to consider radio resources, subscriber history data, usage patterns, mobility patterns, and network health to make UE, access and session management decisions. However, what is still not clear is how such policies can be written. Traditionally, the operator was expected to write rules for making such decisions but that seems impossible now.
It's not humanly possible to imagine every permutation and combination for such vast array of inputs and draft policies. Most vendors continue to reinvent the policy design tools to make such policy design user-friendly. However, what they tend to still miss is the sheer complexity of these policies.
For example, below picture suggests a 5G user usage pattern and Operator is expected to use this information to enforce right QoS/rules without jeopardizing the overall network health.
If you are an Operator, ask your policy vendor to write a real-world network policy. If you are a vendor, try writing one.
This is part of series of blogs for #AutonomousNetworks