Nearly all mobile phone users will testify to experiencing problems with signal fade-ins, fade-outs and break-ups. The same problems can wreak havoc on wireless transmissions of high-speed digital, video or multimedia data. Until now, many scientists believed that the signal fading that causes such interruptions was random, unpredictable and largely unavoidable.
But researchers at North Carolina State University have found that wireless signal fluctuations can be tracked and predicted far ahead of when they occur.
The researchers have developed a forecast technology — an adaptive long-range channel-fading prediction algorithm – said to accurately predict several milliseconds in advance when wireless signals will fade or fluctuate.
‘Several milliseconds may not sound like much advance notice, but it’s the equivalent of hundreds of digital bits in a wireless transmission,’ said Dr Alexandra Duel-Hallen, associate professor of electrical and computer engineering at NC State. ‘That’s enough time to allow a transmitter to switch to other frequencies or antennas that may have stronger signals to the wireless receiver.’
‘If switching frequencies or antennas isn’t possible or practical, the transmitter and receiver would still have enough time to agree to vary the rate of data transmission to minimize the interference,’ said Dr Hans Hallen, assistant professor of physics at NC State.
Wireless signals can be weakened by many sources. These fall into two groups: those that create signal variations over long distances, such as shadowing by buildings or terrain; and those that create strong and abrupt variations in which the signal intensity can change by up to a hundred-fold over distances of less than a foot.
It is this type of signal variation — caused by the overlap of signals reflected from buildings, vehicles, signs and other physical obstacles — that the NC State research team has studied.
The algorithm devised by Duel-Hallen, Hallen and their team not only takes these factors into account, but also accommodates for problems created by high vehicle speeds and high transmission frequencies. ‘In the future, as wireless goes to higher and higher frequencies, this long-range prediction will be essential for efficient adaptation of the transmitted signal to channel conditions due to the much faster channel intensity variations,’ said Duel-Hallen.
To confirm their algorithm’s accuracy, Hallen devised a realistic physical model, and the team tested it against narrowband measurements of real-life wireless signalling from suburban traffic in Stockholm, Sweden.
The data was provided to NC State by Drs Jan-Eric Berg and Henrik Asplund of Ericsson Radio Systems AB. Devising the new model was necessary, said Hallen, because no existing models provided insights into the rapid signal-level variations the team was tracking.
‘The existing models didn’t take into account real-world situations such as the physical movement of phones relative to buildings and terrain, and other factors that can change in a second,’ he said.
Both the forecast algorithm and the realistic physical model technologies are still in the research stages. However, the NC State researchers are talking with scientists at several companies that may further develop the technology for use in commercial wireless phones.