Mobile Boadband Congestion Management & Policy Control


  1. Avatar Chris Horton 28/05/2009 @ 3:28 pm

    When policy causes sessions to be rate-limited or applications blocked, surely that feels like poor service quality to those users? How do networks deal with that?

    • Avatar Randy Fuller 28/05/2009 @ 3:41 pm

      Yes, when sessions are controlled, that user’s performance will suffer in some way. Note that there are some technology details underway that can minimize that–essentially only to limit the user when the network is actually congested in real-time rather than across a broader timespan–the availability of this depends on core & RAN vendors implementation of latest standards.

      In any case, the business question is do all users suffer if there are no selective controls, or certain users according to business rules in the terms of service? There is no free lunch…


  2. Avatar jesper hart-hansen 28/05/2009 @ 3:35 pm

    I am a WiMAX operator:
    Do you calculate the actual capacity utilization in the RAN to understand if congestion is actually occuring in peak hours on specific parts of the RAN?

    • Avatar Randy Fuller 28/05/2009 @ 3:47 pm

      Camiant’s products do not do that calculation–we implement the business rules based on information from other systems.

      Although the ideal solution would be to have real-time congestion signaling, that is impractical to do given the transient nature of mobile packet networks. The current state of the art is to configure times & locations as “peak” based on historical usage patterns, then implement controls based on that. As my earlier post indicated, ideally, the instructions to the network cause controls to be applied only when the packet scheduler is actually congested, but it depends on your network if this can be done. If not, then controls such as IP layer scheduling or policing can be used in most cases.


  3. Avatar Rob S 28/05/2009 @ 4:27 pm

    Are you assuming that the 2% of ‘abusers’ are on ‘unlimited’ plans – do you have any views on billing out of bundle usage as a (belated) control mechanism?

    • Avatar Randy Fuller 28/05/2009 @ 4:57 pm

      The financial model can be configured for unlimited billing plans or for plans that have usage caps–it models a strata of users that consume varying amounts of data regardless. In the example shown, it assumes 2% of users go over a 5GB/month cap and then get controlled, but that can be modified.

      As for how a usage cap actually affects usage patterns, anecdotal evidence from both fixed & mobile broadband operators indicates it does have an effect once a “training” period sets in where the users adjust to loosely understanding a limit, but there are 2 issues:
      1. users do not understand bytes as well as they understand minutes, so the potential of overage charges creates a fear factor that makes the service less attractive
      2. monthly usage caps affect long-term monthly usage, not peak usage which is what actually drives network growth.

      This is why we feel this approach is important, either as a replacement for usage caps or as a supplement.


Leave a comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.