THE Dutch already have more than 10 years' experience with GSM-R after the wayside network was rolled out between 2002 and 2004, followed by a voice system on the train fleet from 2004 to 2006, and the five-year roll-out of GSM-R data for ERTMS on the Betuwe dedicated freight line, the HSL South high-speed line, the new Hanze line, and as part of the upgrading of the busy Amsterdam - Utrecht line.

The Amsterdam - Utrecht line is the pilot project in the Dutch government's plan announced last year to install ETCS Level 2 on the national network. But in 2015 NS will already have to start to renew its onboard GSM-R equipment which should be completed in 2017.

Jos-HoltzerHolzer says that in the 1990s GSM-R was regarded as state-of-the-art with plenty of capacity. It was considered as a communications layer which could be used to provide commercial services to customers, and between 2002 and 2005 NS engaged in joint ventures with private parties to assess its commercial potential for onboard information and entertainment systems. But it found that with a life of 15 years, the total cost of ownership would be double the potential revenue.

In any case the mobile telecommunications revolution means that passengers now have instant access to information and many bring their own entertainment with them. But the consequence of this is that passengers now expect Wi-Fi to be available on trains thereby putting pressure on railways to install it.

Holzer says one of the big problems for railways is that all too often by the time a system is installed it has become obsolescent. He recalled the installation of payphones on Dutch trains in the early 1990s. "By the time we had decided to install them, mobile phones had arrived, and we removed the phones only one year after they had been fitted," Holzer revealed.

"Communication technology regenerates itself every five years, but rolling stock projects take more than five years," Holzer told delegates. The challenge for railways with their slow decision-making processes is how to keep up and avoid what Holzer describes as "today's success rapidly becoming tomorrow's disappointment."

In the Netherlands, GSM-R voice communications became operational from 2005 with virtually 100% radio and safety-specific functions coverage, while GSM-R for ERTMS has been fully functional since 2008. However, NS has observed an increasing number of reliability problems caused by interference. Holzer says the primary source of delays on the Betuwe and HSL lines is the result of communications failure, with interference increasing rapidly in the Rotterdam area from industry and other sources and not just other telecommunications providers as one might expect.

Holzer says these problems are casting serious doubts over GSM-R's data capabilities for ETCS Level 2 even with GPRS. NS fears the problem could get worse with greater encroachment of public providers on the GSM-R frequencies and that the filters currently available are not up to the job of preventing interference. NS has already considered downgrading to ETCS Level 1 to avoid the problem, although this is undesirable operationally, especially at a time when rail traffic is increasing and the demand is growing for signalling and train control systems which can deliver much higher capacity, not less.

Holzer has come up with a radical proposal to move GSM-R away from its current position to a different frequency on the spectrum where it would be less prone to interference. He says it cost less than €500m to roll out GSM-R in the Netherlands, but the revenue from the last frequency auction in December 2012 was a massive €3.8bn. "So why not sell our GSM-R frequencies and use the money to go somewhere else?" Holzer asked. "We need to go for other solutions and fast. This needs to be the subject of serious investigation."

Holzer says the change does not have to be made overnight, but it does need urgent and serious debate. If this does not happen, small densely-populated countries such as the Netherlands may be forced to make some decisions which will mark a major step backwards for ERTMS.