Sunday, June 12, 2011

Are We Facing a Mobile Uplink Problem? Bytemobile Thinks We Do

In the days when P2P was the most popular way of transferring media, the uplink capacity was a very significant problem given the asymmetrical nature of most access networks and the symmetrical nature of the traffic.
A number of [fixed] operators used DPI solutions in order to slowdown P2P uploads only - and the DPI vendors had to understand the "direction” of the traffic, as download flows also generate uplink packets (TCP ACKs, for example) which operators did not want to throttle.

In a recent interview to Mary Lennighan, Editor, Total Telecom, Ronny Haraldsvik (picture), vice president of global marketing at Bytemobile warned that "operators will soon be facing a new challenge: managing the data traffic on the uplink portions of their networks .. End-users will increasingly become broadcasters, uploading video streams and sharing with their friends, That will impact on the uplink that's going to be the next challenge .. As it stands, around two thirds of a mobile radio network's capacity is in the downlink and just one third in the uplink .. Operators will need base stations that are more "uplink-capable", and that's "good news for Ericsson and Alcatel-Lucent"

See "Mobile data: The only way is uplink" - here.

I am not sure I follow the logic, as we need to know first whether people are generally use mobile data services to upload (or do it more from their home fixed connection), and if the ratio of download:upload is less than 2:1, meaning an average of less than 2 views to each video byte uploaded. Note that other applications, totaling to 40% of the capacity (according to Bytemobile itself - here), are also generating primarily download traffic.

Back to the article - "You can't double your capacity by tossing raw bandwidth at it .. For a large operator with a subscriber base of 100 million it would cost $4 billion to add 40% capacity to the network through adding bandwidth alone, Alternatively, the same operator could spend $2 billion to add 20% capacity then put optimisation on top of that, which would leave it with a total capacity increase of 60%. The cost of optimisation would be up to $10 million"

It is not clear how Bytemobile type optimization can help carriers on the uplink, as optimization is placed on the packet core and would not help uploads. I guess Ronny means that with optimization, carriers will be able to set the capacity to a different ratio. Sounds complicated.


  1. You state that optimization in the core does not help uploads. This is an incorrect statement. Server-only optimization can in fact reduce upload bandwidth requirements. A simple example is by causing a reduced number of http requests, total upload traffic is also reduced. There are other examples.

  2. in the case of mobile users that become broadcasters optimization in the core alone will be very limited
    Byte probably has a client solution in mind or in the case of LTE which is all IP: some software solution that will reside in the eNodeB area.
    Not as simple as core only solution but can be done if the motivation is strong enough...