Editor’s Note: Welcome to our weekly Reader Forum section. In an attempt to broaden our interaction with our readers we have created this forum for those with something meaningful to say to the wireless industry. We want to keep this as open as possible, but we maintain some editorial control to keep it free of commercials or attacks. Please send along submissions for this section to our editors at: [email protected].
If you spent $100 billion in one year, what would that buy? It could be Manchester United, the New York Yankees, the Dallas Cowboys and the rest of the world’s 100 most valuable sports teams. It could be 50 nuclear submarines, or 1,000 jumbo jets. But if you expect it to be happy mobile subscribers, you may be out of luck.
Mobile network operators have spent $100 billion worldwide on capital expenditures in each of the last two years to keep up with exploding demand for data services. Yet, in a recent American Customer Satisfaction Index survey, wireless services tied for No. 38 in customer satisfaction among 45 industries. People love their applications, but whenever there’s a hiccup, the subscriber blames their service provider.
Trying to deliver a great experience every time is a daunting task, and high capital expenditures on legacy approaches clearly fall short of completely resolving problems. But new approaches can enhance mobile services by understanding what determines quality of mobile experiences, and taking real-time action as the services are delivered to the subscribers.
User needs and wants
Allowed to make their own bandwidth requests, most mobile applications want a lot. But in terms of perceived quality, some types of applications are much more sensitive to delay or packet loss and may have a very firm minimum bandwidth to have acceptable quality. The radio access network is not inherently application aware and application data is typically carried as HTTP content within the application, thereby receiving default treatment from the network. As a result, some applications are under-served and some are over-delivered in a congested cell.
Streaming video, for example, needs consistent delivery with very low packet loss: research published by the Broadband World Forum indicates that subscribers’ perception of video quality of experience is very sensitive to even a few periods of freezes or pixelation. Certain videos may experience such artifacts when they’re brought down to a lower bit rate or when capacity becomes constrained, and other videos, depending on factors such as their visual dynamics and encoding techniques, might prove more robust in these regards.
Another source of subscriber frustrations during capacity constraints is when web pages download in fits-and-starts, or worse, time out. However, part of the problem may be background application functions grabbing excessive bandwidth from the same cell for a device tucked into the purse or pocket of a subscriber who is not perceiving the QoE.
Capacity is chaotic
Dynamic demand from an array of bandwidth-hungry applications is only half the problem, however, because RAN capacity can be just as complex and ever-changing.
Movement of a subscriber’s device within a cell alters the signal-to-noise dynamics of wireless communications. The RAN and devices will adjust power in real-time to compensate, but these adjustments have resulting impact for other devices too. It’s akin to many people trying to have conversations in a crowded, noisy room. As a result, whenever one device moves farther from the antenna, efforts to maintain that device’s session reduce the respective cell’s overall capacity, and if there is congestion, this also impacts all the other sessions sharing the same cell.
The signal-to-noise ratio for each device in a cell is also impacted by any device entering or leaving that cell. The impact is especially apparent when the device entering or leaving is running an application that may be in the midst of high bit-rate data transfer.
The old protocol guard
So, the mobile industry faces a unique predicament among major types of networks because the fluctuations it faces are both in demand for capacity and in the supply of it. This contention is governed through transmission control protocol.
TCP copes with applications continually seeking to gain more bandwidth. If packets are dropped or corrupted or if there’s a rise in round-trip times from device to network and back, then applications may be forced into a slow-start, in which bandwidth demands are temporarily reduced to recover. When there’s congestion, this results in applications exhibiting a lot of speed variation because of intermittent starts and recoveries.
The mobile network and TCP endeavor for sessions get a proportional fair share of bits. But underlying fluctuations in capacity can wreak havoc with TCP’s coping mechanisms and bandwidth-hungry applications can crowd out other apps. Sometimes the results over-allocate to the wrong types of media while starving important and sensitive sessions.
Operating In The Know
Now is the time to reconsider how mobile networks are operated to better fulfill constantly growing demand. Approaches to date have largely been based on capital expenditures for raw capacity increases and reducing the bit rate profiles of sessions and streams, regardless of their impact on congestion. These remain important exercises. But a new level of quality subscriber experiences is achievable by improving assessment of live traffic flows and by a real-time understanding of contention for cell capacity.
Gaining this network knowledge and acting on it require best practices from a variety` of disciplines, including wireless transmission, packet-based networking, and digital video processing. Fortunately, constant technological progress is enabling new possibilities for effective interoperation across these fields.
This includes knowing which cells do and do not face imminent congestion. It means keeping tabs on each session’s bit rates across various applications, including sensitivity of experience quality across speeds. It demands insight into each cell’s conditions and contention between subscribers’ applications during congestion. It is important to be able to see roaming subscribers’ traffic as well since these subscribers produce a lot of revenue and are in contention for bandwidth.
When there is no congestion, it is generally best to remain hands-off so traffic flows proceed with maximum quality. When congestion does occur, it is important to detect it and take action in the moment. A recent 3GPP study determined that 90% of congestion events last less than two seconds. Reaction time is crucial – too slow and the damage is already done.
Operators that can incorporate this detection of congestion and real-time management of application traffic into their networks can maximize revenues and rein-in capital expenditures. Most importantly, they’ll be able to deliver the experiences that will make each subscriber a rave reviewer.
