It’s the Signaling Stupid!
July 22, 2012
If you have been following the wireless industry, you for sure would have heard of the capacity crunch and exploding demand for mobile traffic data services. Hardly any presentation by a vendor, analyst, and even operator starts without a graph of future traffic trends. But quietly, another problem has been brewing; one that has received little media attention, but which operators, vendors and industry insiders have recognized as a real and serious issue that warrants much work to contain: it’s the problem of exploding signaling traffic on mobile networks. To illustrate the challenge, Telus reports that in Canada while the data traffic has seen tremendous growth of over 100% year-over-year for the past three years, the signaling traffic has grown by a staggering 2700%. Deutsche Telekom reported 360% increase in signaling traffic over a two year period spanning 2010 and 2011 even with the introduction of techniques that reduce the signaling load.
The root cause of the problem is a function of how the network and mobile devices were designed to operate and the traffic pattern of data services. To save battery power, mobile devices frequently change states between active and idle modes. This works well for voice services when a mobile is in idle mode unless a call is in progress. But that’s not the way data services work. Data usage is intermittent and consists of several short sessions which results in many network connect/disconnect events. Applications such as email, Facebook or Twitter operate by poling the network for updates. As they are un-coordinated and lack knowledge of network conditions, they lead to a large amount of signaling.
The process to set up a radio channel to transmit and receive a small amount of data is rather inefficient, adds latency and results in reduced cell capacity and performance: in 3G, it takes about thirty messages to go from idle to active modes for mobile initiated transmission. When the network wants to push information to a device, the process is longer as the device need to be paged and located first. To improve the process, enhancements such as network-controlled fast dormancy, Cell_PCH and Cell_URA techniques have been developed to ameliorate the situation although they do not provide a solution (they add two intermediate stages between idle and active modes that require communication with the RNC at each stage and it is possible for devices to circumvent operator’s settings targeted at reducing signaling load).
The introduction of LTE can help moderate the signaling problem but is not expected to solve it. LTE introduces enhancements to the radio interface, implements a flat IP architecture, and separates the control and bearer planes. Between 7 and 13 messages will be required to go from idle to active mode (as opposed to about 30 in 3G) and between 4 and 6 messages are needed to go from active to idle mode. Such enhancements will have a positive effect that includes reduced latency in addition to reduction in signaling, but the greater types of commercial services will add to the signaling requirements.
The impact of the signaling problem is not just limited to the cell, propagates through many network elements and has a strong impact of the design of Radio Network Controllers (RNCs) in 3G networks as well as the gateways and network elements comprising the LTE core network (e.g. MME, S-GW). Telecom vendors can differentiate their solutions by how they provision to tackle the signaling problem: the architecture and design of core network elements is another aspect of tackling the signaling problem.
Another factor that can help reduce the signaling load includes the operating system of the mobile device. For example, Apple iOS and RIM push notification service interacts much better with the network while Android remains a challenge.
The design of mobile applications is another front to reduce signaling. The GSMA has recognized the significance of this problem and announced an initiative to create more efficient mobile applications at this year’s Mobile World Congress.
In considering the signaling load, we need to recognize that enhancements to the wireless network such as WiFi offload integration and small cell base stations will also add to the signaling traffic. Many Machine-to-Machine (M2M) applications are also notorious for having a traffic profile that consists of intermittent small amount of data. How to deal effectively with these issues will be indicative to the success and tractions such solutions and applications will have.
I think the signaling problem will have a prominent role in affecting operator’s strategic decisions on future investments in 3G networks, LTE roll out strategy, migration of 3G subscribers to LTE, and spectrum re-farming decisions. The consequences are far reaching and will be interesting to see them played out in the years to come.