Volume not the only cause of wireless network stress
Data is now stressing wireless networks in ways that have never been experienced before, as improved devices, better 3G networks and a greater number of richer applications are available to users.
But according to Michael Schabel, general manager of 9900 WNG at Alcatel-Lucent Ventures, applications that might seem more likely to place greater stress on mobile networks may not be as bad as you think, while others are much worse than you would have expected.
Although networks are far from saturated, the "deluge of data" in the past 18 months is starting to put stress on networks in different ways, depending on what application or device is being used where, said Schabel.
He added that there is now already evidence that customer experience is being degraded. The problem is how to identify what is causing network performance to deteriorate, and Schabel said the causes range from problems with the device being used through to buggy applications and viruses. Although the network operator is usually the first point of contact for a user, the network itself is often not the problem, he added.
"The user will always blame the operator and call them first," he said. But sometimes it can take months to locate what the problem might be. The result of this is that it's becoming increasingly complex and expensive to deal with customer care.
The temptation is always to throw extra capacity at problems such as a poor connection, slow and variable speeds and inconsistent quality of service. But Schabel said network management is really the key here, to better manage network contention. For example, sometimes the answer might be to modify the network to accommodate a certain application that is causing contention in one particular cell.
Schabel said users often experience problems because of "localised congestion" in a cell, for example. Defective devices can often make network access appear slow, while an application that is constantly making a request to the network can also slow speeds down.
"It's a cell-dependent problem," added Schabel. "The network needs to be managed down to the cell…each cell is different."
For example, an analysis of each cell will show a different mix of applications that are currently in use, from email through to online games. And every application makes use of wireless network resources in different ways. Some are very inefficient and keep the device on the network longer than it needs to be, which also uses up more battery life. Some applications use more signalling resources than others, while others simply push a high volume of data over the network.
Schabel said the volume of data as well as how much use is made of signalling and wireless resources all play a role in dictating what the end-user experience will be.
But he stressed that the applications often believed to provide the most problems for networks, such as peer-to-peer or video streaming, are actually "the most efficient on wireless networks that I have seen. I'm not saying there are no problems, but I have not measured a more efficient application" than peer-to-peer, he said.
In fact Schabel contends that applications such as push email and location-based services are far more inefficient and cause more network problems. This is because unlike streaming or P2P, email is unpredictable and is constantly making demands on the network that are not always actually necessary. "Both video streaming and P2P are very predictable," he said. When there is a problem with how an application uses a wireless network, "more often than not it's not a volume issue," he said, but is more to do with how efficiently an application uses signalling and wireless resources.
Alcatel-Lucent's solution to the problem is the 9900 Wireless Network Guardian that detects what is running on the network to help operators resolve network problems. The detectors sit in the packet core and help build a real-time model of the network. The network load is then inferred from the information provided.
"It identifies the 'expensive' traffic on the network," said Schabel. "We look at every packet and signalling in real time…we can identify what is hurting the network in real-time."
Источник: Total Telecom
Заметили неточность или опечатку в тексте? Выделите её мышкой и нажмите: Ctrl + Enter. Спасибо!