A signal too noisy: Operators at risk of signaling traffic deluge
Under the massive growth in data traffic, mobile networks are grinding to a halt. Mobile operators are frantically trying to increase capacity for the smartphones already in use and those that they expect to sell soon, along with new devices such as the iPad or the Kindle--and eventually machine-to-machine terminal devices. More base stations and fatter backhaul pipes are the key lines of intervention. More recently, mobile operators have even started to hint that data usage caps may soon replace unlimited plans to keep traffic growth under control.
But is data traffic the only--or main--cause for the congestion in mobile networks? A new suspect has emerged: the increase in signaling traffic, primarily due to data-centric, always-on devices like smartphones, which keep connecting and disconnecting from the cellular network to prolong their battery life, to keep their owners in touch with their friends through email, messaging and social networks, and to maintain applications active in the background.
Intrigued by the possibility that managing the devices may take more resources--and sophistication--than actually transferring data itself, I talked to what seemed like countless people and things got even more interesting. Nobody seems to agree as to what impact signaling is having on traffic today or that it will have in the future, with the increased adoption of IP-based technologies like WiMAX or LTE--or how to address it.
Some think that while signaling traffic is obviously increasing, cellular networks can manage it. Others think that it can be a bigger challenge that data traffic itself--one that has started to impact the ability of cellular networks to carry voice traffic as well. Understandably, cellular operators themselves have been reluctant to share data, and vendors have an interest in pushing their solutions that increase signaling capacity.
It is still unclear how much of the congestion we see today is due to signaling and not to data traffic, but the increase in signaling traffic is a genuine issue that will only worsen with the increase in smartphone penetration and the mass market adoption of new devices. In a simple world were feature phones dominate, signaling is a boring, background activity needed to manage a few phone calls per device that adds little overhead to the network resources.
Data-centric devices like smartphones have very different signaling requirements. To keep email updated, BlackBerry phones need to establish and tear down connections to the cellular network very frequently. The connections are short lived and carry very limited amounts of traffic, but generate a higher ratio of signaling to data traffic than, for instance, large downloads on laptops.
Furthermore, BlackBerry devices and other smartphones establish frequent but short connections to preserve their stand-by battery life. They establish a connection, check email, discontinue the connection and go into a sleep mode until they need to check email again. Or, the email server will wake up the device, establish a connection, send the email, close the connection and let the device go back to its sleep mode. To device manufacturers long battery life is a more valuable than saving network resources, and until now even mobile operators were not concerned about the additional traffic generated by this behavior because it was a small portion of the overall traffic.
Email checking is responsible for a rapidly decreasing portion of signaling traffic. Social networking, location based, and VoIP applications are increasingly responsible for substantial background activity that require, like email updates, frequent, short connections to keep subscribers in constant touch with their family, friends and colleagues. In addition to signaling traffic, they also generate additional data traffic that may add up in unpleasant ways when roaming internationally, or when the subscriber has a plan with a low traffic cap.
The increased signaling traffic exerts some pressure on the already congested RAN, but the main pressure point is the core network, which lacks the resources to deal with the increase in signaling traffic. When signaling traffic exceeds capacity, congestion ensues. In GSM and HSPA networks, where the same channel transports both voice and data, congestion affects both voice and data services. This is particularly worrisome to operators as their profits come largely from voice and not data (data revenues are clearly increasing, but data is not nearly as profitable as voice). Subscribers unable to place a call, sending a message or trying to watch a YouTube video, will keep trying, further increasing congestion.
Will WiMAX and LTE solve the problem? It is unlikely, and in fact they may make matters worse as WiMAX and LTE devices are likely to generate even more signaling traffic, because of a more intensive use of VoIP and of applications with substantial background processes. Technologies like WiMAX and LTE will also promote the adoption of much higher number of terminal devices--e.g., consumer electronic devices like cameras or digital players, automotive modules, mobile health care devices, and M2M devices such as smart electricity meters--that may generate little data traffic, but comparatively high levels of signaling.
The good news is that the rise in signaling traffic is easier to manage than the rise in overall data traffic, but action needs to be taken soon to contain congestion problems as smartphone (and smartphone applications) adoption increases.
The simple, brute-force approach of installing more equipment will not suffice. Mobile operators need to develop a new approach to signaling which requires collaboration with device, infrastructure and core network vendors, and OS and application developers to optimize how signaling uses the network resources.
Cellular operators are going through a fast learning phase that will give them a better understanding of how signaling impacts their network, and of how to manage it more efficiently. They will clearly need to increase the signaling capacity in the core network to prevent congestion, and the inefficient use of network resources it generates. At the same time, devices, OSs and applications have to become more aware of the impact that signaling has on cellular networks, and develop a more thrifty approach to transmission. It is unlikely that device vendors and applications developers will lead the change because they are only marginally affected by signaling traffic. Mobile operators will have to take the lead. And a useful first step could be to acknowledge more openly the smartphone signaling overload--if signaling is indeed a real issue.
Note: I am particularly indebted to Woojune Kim at Airvana and Manish Singh at Continuous Computing for sharing their insight on how signaling is evolving with the introduction of data devices and IP-based networks. Of course, the article does not represent their opinions and I am the only one responsible for any inaccuracy.
Monica Paolini is the founder and president of Senza Fili Consulting and can be contacted at firstname.lastname@example.org. Senza Fili Consulting provides expert advisory services on wireless data technologies and services.