Next-gen cellular nets need more expertise

At one time, the technologies designers embedded in test, measurement and monitoring equipment sometimes seemed to lag the technologies engineers were developing and deploying. If this ever actually was the case, it certainly no longer applies, especially in the increasingly heady world of cellular networks.

Engineers face novel, concurrent challenges in the deployment of second- and third-generation mobile wireless networks. Difficulties involving high-speed data links, network interoperability, complex signaling layers and the delivery of Internet content to mobile handsets all require attention.

Deploying and managing those networks requires expertise in circuit-switched voice, packet data, Internet protocols and applications, new network elements and more. In these belt-tightened times, operators can ill afford to staff field offices with experts in all things, so designers need to come up with other enabling solutions for test and troubleshooting.

The upgrade paths to 2.5G and 3G services differ, depending on the technology starting point, but all carriers follow a general pattern. Global System for Mobile Communications (GSM) operators and time-division multiple access operators that are migrating to GSM begin adding packet-handling capabilities at their basestations and in the core network, creating General Packet Radio Service networks with demonstrated data rates up to 115 kbits/second. CdmaOne operators did the same, simultaneously implementing an upgrade to the code-division multiple access (CDMA) air interface, creating cdma2000 1xMC (1 times multichannel) networks with data rates up to 307 kbits/s.

Regardless of the specific technology path engineers use to achieve high-speed mobile wireless multimedia service, the new complexities facing the network engineer are similar. One common change is the migration of problems from physical and data-link layers into the domain of the application, which means that more and more trouble tickets will have to be resolved via techniques akin to software debugging.

In 3G networks, data sessions use Internet Protocol (IP) for transport and, unlike a circuit-switched call, do not rely on the nailing up of a dedicated circuit across the network. In order to deliver all the content of the Internet to mobile wireless roaming users, engineers established complex sets of new protocols riding on top of IP. While problems such as packet loss or latency can certainly still occur in IP-based transport, the complexity and troubleshooting challenges lie primarily in the new software-implemented protocols.

For example, consider some protocols involved with communications through the radio network controller (RNC) in a Universal Mobile Telecommunications System network. There are actually more protocol layers here than in the standard, seven-layer OSI model. Thus, this classic model is no longer sufficient for representing the complexity of 3G networks.

In addition, protocols on either side of the network element are changing. That translation does not lend itself to a one-to-one mapping, rendering traditional troubleshooting techniques such as call tracing very problematic.

Another area of change is the number of signaling messages that must be found and monitored, often referred to as the dynamic logical link assignment. In older SS7/GSM networks, a single signaling link carries the signaling messages of approximately 1,000 users. The link is permanent and is easy to find and monitor.

In contrast, 3G wideband-CDMA technology has up to six logical signaling links for each user, which do not necessarily reside on the same physical link. As a result, up to 6,000 logical links may have to be monitored in order to track the signaling messages of 1,000 users, leading to the requirement for autoconfiguration of local links.

Clearly, the telecom industry needs a new class of monitoring and test solutions capable of handling the new complexities. But simply handling the testing of new complex protocols and interfaces is not sufficient. Communications test solutions as complex as the communications themselves put the burden of multitechnology expertise squarely on the shoulders of the engineer. The next generation of deployment and troubleshooting tools must provide expert analysis of network conditions and potential trouble and make complex tasks transparent to the user.

Next-generation test tools must be able to handle the vastly increased volume of control messages. That means new tools must be able to capture line rate data. It also means the tools must know how to capture and analyze the data in real-time, since network troubleshooting is a real-time activity. Because many faults are intermittent, however, the test solution simultaneously should be able to capture, display, analyze and store the same data stream.

The test solution also must possess the built-in intelligence, via expert system software or the equivalent, to set up complex tests, sift through vast amounts of data to target network trouble and then drill down to the relevant data to the root-cause problems. For example, next-generation test gear should provide autoconfiguration of specific interfaces. Such a feature greatly reduces time to first measurement and mean time to repair.

Last but not least, with the addition of many new network elements and interfaces comes the problem of fault isolation. To troubleshoot specific service problems in next-generation networks, engineers must be able to track the progress of sessions across the different elements and interfaces.

This multi-interface tracing is a prerequisite for effective network and service diagnosis. To enable it, test tools should have parameters to identify the subscriber's identity and to trigger a search. That allows the engineer to conduct the in-depth troubleshooting necessary to locate the problem's root cause and view all details related to the protocol messages.

Wayne Newitts is marketing manager for the Americas and Ralf Kreher is monitoring and protocol test product manager at Tektronix Inc. (Beaverton, Ore.).