Predicting Network Performance Can Save Millions
Organizations looking to invest millions of dollars in a new communications network would be wise to develop a digital model of the system before they commit to major hardware purchases. Although this advice seems common sense, an expert said, it is surprising how many companies and government agencies agree to buy new networks without really knowing if they will work properly and meet their needs.
“When you are spending hundreds of millions of dollars on a system, spending a small percentage on a simulation can help avoid making a large mistake,” said Robert F. Brammer, chief technology office at TASC, a division of Northrop Grumman Corp.
The company specializes in government and defense-related information systems and analysis services. It has spent corporate research dollars recently to come up with modeling technologies to simulate communications networks and signals processing.
“There are several U.S. federal customers in sight” for this technology, Brammer said in an interview. Projects currently underway include communications network models for commercial satellite firms, he said. The modeling covers the network’s basic physical layer—predicting signal-to-noise ratios and achievable data rates in various types of environments. The communications take place via the Internet, by wireless terrestrial channels and by space-to-ground channels.
“We worry about the fundamental physical communications channel and network issues [such as] computer-to-computer protocols and information security” that have to be included in the model, said Brammer.
Models are done for different purposes, he explained. “In some cases, you want to be able to predict the performance of a communications system or network that a customer wants to build.” In other instances, customers want to predict the network’s performance in various environmental conditions.
“In other cases, they already have a network, so we build a detailed model. In the event of an outage or emergency, we can help with tactical planning of how to restore services,” Brammer said. “There are number of reasons for the simulation models.”
High-fidelity models can predict failures, such as bottlenecks in transmission, bandwidth availability and whether there is enough computational power, he said. “They may be interested in trading performance vs. security.” That means they must decide how to transmit information at different levels of security through a network. “You have to make sure that only the people who are authorized to get the information can get it quickly and easily,” said Brammer. “That is easy to say, but very difficult to do in practice.”
Modeling of computer attacks is done frequently, he said. “Some of these attacks have been well coordinated, involving thousands of computers. So we have to be able to model what might happen and what to do about it.”
The challenge for the industry, he said, is to “understand how threats to the performance of communications systems are changing as a result of rapid changes in computing and communications technology.” At the same time, he added, “a lot of the technology is so new and relatively immature, so there are security issues that you have to be careful about.”
In information security, Brammer said, “we have a lot of learning to do.”