The Big Picture

What do a soccer mom, a teenager and a road warrior have in common? They're all wireless subscribers. But each uses wireless differently, and the more you can understand and anticipate their habits, the better you can engineer your SS7 network.

"It has nothing to do with SS7, but it helps you to model your system and understand what sort of traffic patterns will be generated by certain events," said Steve Racz, Nortel senior manager of GSM signaling development.

Marketing forecasts can be a valuable tool for determining how much traffic the SS7 network will have to accommodate. Before debuting a new rate plan, it's important to understand what types of users it will attract.

"Nationwide programs such as AT&T's Digital One Rate are driving a lot more roaming air minutes than we've seen in the past," said Tony Janssen, Nortel GSM product-marketing manager. "That creates more demand on the SS7 side. We have a lot of operators that are already doing advanced planning for flat-based minutes of use. Ten cents a minute or even sub-10cents is really going to alter their traffic patterns."

But anticipating where and how those new subscribers will use their phones provides only half the picture. Subscriber usage patterns change as they become familiar with their phones, and the more they use them, the greater the SS7 workload.

"Without understanding what your subscribers' profiles are, you don't have any reliable way to predict how your network is going to react," said Robert Gentry, Nortel GSM product-line manager. "We've seen dramatically different profiles from different customers."

FCC mandates are another guide for gauging SS7 capacity, but they don't preclude marketing forecasts. Case in point: E-911. Most carriers are eyeing enhanced services that use the E-911 location technology as a way to defray implementation costs, so the SS7 network must be able to handle not only distress calls but also normal use of the enhanced services.

"Every one of these new applications and requirements that are coming out requires new application messages riding on SS7," said Randy Snyder, Synacom Technology director of systems engineering. "It's just more traffic that these networks have to carry. They need to look to the future for additional applications, subscriber growth, type of traffic and the size of the messages."

Those forecasts help ensure that SS7 supply will meet demand.

"If you've got the opportunity to do 5-year planning or even 18-month planning, that would give you the tools to reduce your SS7 messaging traffic significantly," Gentry said. "The key to avoiding all these congestion problems and traffic problems is network design."

It's easy to get lulled into a false sense of security because the steady increase in the size and number of SS7 messages isn't clogging up networks -- yet.

"Network providers look at that as a potential in the future," said Dave Cox, Tekelec director of product marketing for the network-switching division. "Right now, it's not a problem, but who knows what the new services are going to bring down the road?"

MOVING TARGETSWireless is more demanding of SS7 than wireline is.

"The main thing that drives SS7 in wireless is the mobility of subscribers," Racz said. "That's unpredictable, and it puts a lot of pressure on the engineering of the SS7 backbone."

Roaming is the ultimate form of mobility. One way to reduce roaming's SS7 workload is Nortel's Supercharger, which caches subscriber data within VLRs that the subscriber frequently visits.

"When a subscriber moves from one MSC to another, the subscriber data doesn't necessarily have to be retrieved from the HLR a second time," Racz said. "If it quickly moves back into an area that it just moved out of, then the data that are there are re-used, and that cuts down on the traffic."

Maintenance also can cause short-term congestion.

"Sometimes you'll have events that occur that seem unrelated to where the congestion actually takes place," Racz said. "For example, they'll take down a cell site, re-engineer it and put it back up again. Now everyone in that cell site will try to re-register, and so you have an influx of messages coming from the radio into the network. That itself may not be a problem, but then, as all the messages concentrate toward the HLR, you may find that there's a bottleneck on the HLR side."

Global-title translation also can affect performance.

"Every time you have to look at one more digit of a number called or calling-party number, that's a performance hit because you're literally looking at millions of messages a minute," said Synacom's Snyder. "Make sure that if you have to do digit analysis on one more digit, you're not reducing the performance of the network. Does the SS7 network provide for different types of global-title-translation addressing? What we see is a lot of STPs (signal transfer points) taking longer than they should to implement a new type of global-title address."

FLEXIBILITY FOR THE FUTUREIt's impossible to predict every type of subscriber and service that will appear, so choosing an SS7-network architecture that provides the most flexible platform for growth is another way to ensure efficiency and reliability.

"It's going to get to the point that a single SCP (service control point) per network is not going to cut it anymore," said Dan Bantukul, senior manager of product marketing for Tekelec's Intelligent Network Diagnostics division. "The SCP has to be distributed across the network. You cannot have a single SCP doing the load of the whole network. Distributed SCP architecture is the architecture of the future."

The distributed approach should be a particularly good fit for smaller carriers because upgrading will mean adding more elements instead of replacing them.

"Distributed architecture should work well for the smaller operator because they can start small and grow as they need it," Bantukul said. "If you start with a central-SCP architecture, when your requirements grow, you're going to have to do a 'forklift upgrade' of your SCP to be able to handle the load. But if you go with a distributed architecture, it grows as your needs grow without having to do a forklift upgrade."

Regular SS7-network surveillance and testing are the best ways to monitor performance and determine where capacity is stretched thin.

"Use an SS7 emulator or monitor to look at the signaling network for the results," said Kenneth Ross, Coastal Technologies senior signaling consultant. "The monitor/emulator is also useful to introduce a new signaling protocol or parameter to the network and see its effect prior to a full network release."

One way to maximize the investment in surveillance equipment is by providing all departments with the data collected.

"Using the signaling data to construct a picture of the underlying traffic flow results in a near instantaneous picture of network overflow, fraudulent activity and customer-account usage," Ross said. "When deploying (SS7) probes, you should ensure that they can forward signaling messages to a third-party processing platform. This allows for a more detailed analysis of the signaling data and can drive most of the network management, customer support and even marketing requirements of a telco. (It also) allows for coverage of unusual requirements, such as managing specialized hybrid network elements or handoff to competitors and partners for local access and termination."

One important statistic to monitor is link loading. Links are engineered to run reliably at .8 Erlangs, but .4 is the real limit because if another link failed, the other links would have to handle the additional traffic. A safe range is .2 to .3 Erlangs during the busiest part of the day.

"At .4, you'd begin to plan for adding additional links," said Nortel's Racz.

Monitoring the links' traffic loads is one way to identify if upgrades are needed, but monitoring the overall quality of service (QoS) means looking at more than just link loads and database query-and-response ratios.

"As an industry, we have to shift the paradigm from looking at the raw SS7 information to the services enabled by SS7," Bantukul said. "If you get a million good responses back, does it mean your 800 service is working at 100%? No, because the 800 service interworked with the call setup, which is another part of SS7. So if you look at the 800 query to the database alone, it doesn't gauge how well your service is doing. It only gives you how well your 800 database is doing."

Assessing QoS means looking at other issues, such as trunk setup and call setup.

"Every service is now interrelated, so looking at part of the database alone will not give you the total picture," Bantukul said. "The only alarm mechanism you have to tell you something is wrong with the service is the subscriber. If SS7 works well, they're assuming that the service will work well also. That's not the case."