 | Issue: | AME 2012 |
| Article no.: | 4 | |
| Topic: | Mobile congestion management is key for Africa | |
| Author: | Tom Donnelly | |
| Title: | Co-Founder and Chief Operating Officer, Sales & Global Services | |
| Organisation: | Sandvine | |
| PDF size: | 228KB |
About author
Tom Donnelly is Co-Founder and Chief Operating Officer, Sales & Global Services of Sandvine. Mr Donnelly is responsible for worldwide sales, support, professional services and business development at Sandvine.
An accomplished telecommunications entrepreneur, Mr Donnelly was part of the executive team for PixStream – which was later sold to Cisco Systems for more than US$500 million. He was responsible for sales and business development, winning prestigious customers such as Bell Canada, Eircom (formerly Telecom Ireland), Telenor and SingTel, as well as building a successful partnership and channel program that included Newbridge Networks (Alcatel), Nortel Networks, Next Level Communications (Motorola) and Cisco Systems.
Mr Donnelly’s international experience includes executive-level positions managing marketing and business development operations in the Asia Pacific Region, Europe, Latin America and North America. He has been involved in the telecommunications sector since 1990 and is a popular speaker and recognised authority on broadband market issues. He has spoken at notable events such as Broadband World Forum, Supercomm and CommunicAsia.In 2007, Mr Donnelly and his fellow co-founders were honoured with the Ernst and Young Entrepreneur of the Year award in the technology category for their success and dedication to their work at Sandvine.
Tom Donnelly holds a Bachelor of Arts degree from McGill University and is a graduate of the Wharton executive development program at the University of Pennsylvania.
Article abstract
To respond to the surge in demand for mobile communications while preserving reasonable quality and reduce infrastructure costs, operators must manage mobile Internet congestion. The, traffic management cannot be static. It must isolate the congested resource, the application that needs protecting and the users who are consuming more than their fair share. Sandvine’s own research shows that quota management is not the best solution, and high volume ‘long-term’ consumers may not be the cause of congestion. Instead, it is important to constantly monitor the network and pbtain dynamic resource-level and user-level visibility. Awareness of the application type allows prioritising ‘experience now’ traffic (such as HTTP and VoIP) over ‘experience later’ traffic (email, bulk and P2P downloads).
Full Article
The increasing availability of mobile broadband in Africa is beginning to fulfil the promise of communications and worldwide connectivity for African subscribers. As communications service providers deploy mobile networks, Network Policy Control technology such as congestion management will boost the cost-effectiveness of data delivery while ensuring that subscribers enjoy a quality service they can rely on. Service providers and subscribers alike will benefit from advanced capabilities such as subscriber mobility awareness, dynamic congestion detection, and application-aware prioritisation.
The GSM Association’s annual African Mobile Observatory report shows that in September 2011 mobile penetration in Africa reached 620 million subscribers and is expected to grow to 735 million subscribers by the end of 2012. The report notes that, while mobile penetration is high, fixed-line service penetration is very low and is likely to remain that way:
Of particular note are countries such as Cote d’Ivoire and Benin which have 95 per cent and 87 per cent mobile penetration, but less than one per cent and two per cent fixed-line penetration respectively. Even in the more developed countries in Africa, such as Algeria and South Africa, fixed-line penetration remains below ten per cent. In most of the countries of the region, fixed-line penetration is unlikely to grow further as this would require extensive investments in infrastructure. Consequently, the mobile phone will remain the predominant means of communication for the majority of people.
GSMA analysts say Africans have shifted to mobile communications because landline service is unreliable and too expensive. Although 96 per cent of these mobile subscriptions are pre-paid with voice services currently dominating, data subscriptions are steadily increasing. This means for the vast majority of African subscribers, their first internet experience will most likely be over a 3G or 4G mobile connections. Subscriber access methods will include smartphones, tethering and mobile USB sticks attached to PCs and laptops. For mobile operators, one of the biggest challenges is managing congestion in the radio access network. Regardless of how or where people connect to the mobile network, subscribers are looking for the best possible quality of experience.
What is network congestion?
Sandvine defines congestion as follows: “Defined on a per-application category basis as the variability in delay or packet loss beyond what the application can withstand without the user noticing”. In other words, congestion is something caused by packet delay and latency, as framed by the user experience – you know it when you see it. A situation of congestion which causes an instant message to be delivered 500ms later is irrelevant to the average user, but the same delay on a VoIP packet could mean a garbled conversation that causes a negative user experience. In practice, network congestion tends to be both short-lived (most congestion occurs during peak period or due to ‘events’) and very localised.
What causes network congestion?
Studies on both fixed and mobile networks show that a minority of subscribers consume the majority of bandwidth. In one case, Sandvine has seen five per cent of subscribers consuming 80 per cent of the network’s bandwidth over a monthly period. Additional studies looking at the factors affecting network congestion have concluded that heavy ‘long-term’ consumers are not necessarily the main contributors to congestion. Sandvine’s own studies clearly show that when a cell is congested, a minority of heavy ‘short-term users’ (15 minute intervals) are consuming the majority of bandwidth. This can be anyone at any time – we all contribute disproportionately to network congestion at some point in our weekly routines.
Mobile network congestion patterns are more complex, and less predictable. For example, the medium is affected by environmental factors and a cell’s maximum capacity can vary from day to day – static thresholds are inaccurate in this case, making it more difficult to reliably preserve infrastructure and predict capacity expansions. It’s also much more expensive to deliver mobile data as compared to fixed.
Congestion affects user perception and experience, and in that sense the application type can be an important factor. ‘Experience-later’ applications such as P2P downloading can greatly impact ‘experience-now’ applications such as video streaming and VoIP if they dominate during congestion periods.
What is advanced access network congestion management?
Network Policy Control presents a more attractive alternative to service providers for addressing mobile congestion than buying more capacity equipment. It’s also beneficial to subscribers, since building out capacity does not guarantee an improvement in quality for the majority – the heaviest users will always make use of available bandwidth.
The most accurate, auditable, fair and ultimately effective solutions are those that target congestion alleviation in the access network. When minimising the impact of congestion through Network Policy Control, what’s required is a solution that looks for, identifies, isolates and manages congestion based on the true causal factors while delivering maximum quality of experience to the maximum number of subscribers. In practical terms, such solutions let operators remain profitable while still allowing subscribers to actually benefit from today’s amazing Internet innovations.
More often than not, congestion management is about Congestion Minimization to ensure that only short-term congestion contributors are affected, and only at times of congestion. This minimisation is usually accomplished by decreasing the amount of low priority traffic, so that the high priority traffic has more capacity. The ‘High’ and ‘Low’ can be defined as normal vs. short-term heavy users, or real-time applications vs. non-real-time, or a combination of both.
What can we learn from African broadband usage today?
As the GSMA report indicates, very few African subscribers are currently accessing broadband internet over mobile, so it is difficult to derive meaningful statistics about mobile application usage. Sandvine’s Fall 2011 Broadband Phenomena report included a spotlight on emerging countries with detailed statistics about fixed line application usage in Africa. This information can be a useful guide for projecting mobile behaviours, since Sandvine has repeatedly seen fixed-line behaviour mirrored in the mobile space when examining detailed application trends across the globe.
The spotlight on Africa showed that, unlike Western and Eastern European or South American application usage, in Africa P2P dominates upstream bandwidth at 44 per cent. In terms of downstream bandwidth, unlike in North America where web browsing accounts for 17 per cent, P2P 7 per cent and real-time entertainment over 50 per cent, an analysis of Africa shows web browsing at 35 per cent, and real-time entertainment and P2P equal at 25 per cent each.
What does this tell us? It suggests that a majority of African mobile internet users accessing real-time applications during network peak and busy times may find their experience frustrated by a small minority of subscribers generating the majority of the traffic.
Fair Access, Subscriber Mobility Awareness, and Dynamic Detection
“Fairness” emerges as the critical subject of focus in the consumer-operator discussion about best practices for network congestion management. Who should be affected by traffic management when the network is congested? Where should traffic management be applied, and based on what criteria? Which applications should be impacted and which should be protected?
In mobile networks, to isolate the true contributors to congestion, operators must use real-time, cell-level traffic management in the radio access network. Cell-awareness is important to achieve resource-level visibility and, like access network segments in the fixed space, cells don’t move. However, subscribers do move, and a congestion management approach that doesn’t account for this reality is fundamentally flawed.
The most effective solutions are those which detect and isolate the congestion problem whenever and wherever it occurs by monitoring cell congestion levels and Subscriber Access Location in real time. An access network congestion management solution with subscriber mobility awareness allows operators to ensure traffic management is only applied to short-term heavy users currently attached to congested cells.
Without mobility awareness, operators and subscribers are forced to settle for a bandwidth management policy that is inaccurately applied to those who are not causing congestion, or which targets aggregate traffic without focusing on the contributors to congestion at all. Just as bad, the congested resource is not properly isolated and the reduced efficiency results in a poorer preservation of access network resources and a lower return on investment from the Network Policy Control solution.
With fairness and precision established, layer-7 application-awareness adds a powerful third dimension to mobile congestion management policies. During times of congestion, subscriber quality of experience can be further protected by prioritising ‘experience now’ subscriber traffic (such as HTTP and VoIP) over ‘experience later’ traffic (such as email, bulk and P2P downloads). Traffic Management systems must accurately track subscriber location and cell traffic quality throughout the access network in real time, and automatically apply this awareness to a dynamic congestion management policy without an operator’s intervention.
The inescapable conclusion is that the more accurately you can target the actual cause of mobile network congestion, the better you can protect subscriber quality of experience during times of congestion. This is where subscriber and operator interests meet – better congestion management delivers greater cost-savings to keep the delivery of mobile internet services profitable, while ensuring African subscribers have a good broadband experience in the years to come.
