ERA’s proposed specification for modulation was called GMSK (Gaussian minimum shift keying), while the DFE referred to earlier was used as an equalization method.
”We had begun to get our equalization technology to work and it functioned reasonably well in Paris,” says Jan Uddenfeldt. Afterwards, ERA’s people tested Viterbi’s algorithm on their own equalizer in the DMS90. ”We were able to see that its performance improved considerably. We presented the results at the IEEE (Institute of Electrical and Electronics Engineers) conference in May 1987.”
Östen Mäkitalo presented his 8PSK solution at the same IEEE conference. When the American and Japanese competitors to GSM were launched a few years later, he noticed that both systems were based on versions of the 8PSK. Even later GSM/EDGE was also to incorporate 8PSK.
Mäkitalo’s colleague Mats Nilsson, who was in charge at the Paris tests, says that in practice it was Televerket’s MAX-II that led to the reduction of the TDMA factor (the “time windows”) in the GSM standard from 10–12 to eight. “This was decided at something that resembled a farmyard auction during a long rainy meeting in the Hague in February 1987. ‘Is there any support for five channels ... no ... perhaps six channels ... not really ... well eight then?’ Fortunately there was no support for the idea of going as low as four.”
Televerket Radio’s staff at that time included Håkan Eriksson.
“At this stage, industrial members were not admitted by CEPT, which was after all an association of European operators. On the other hand, each national delegation was allowed to take along two industrial experts to each meeting. I was involved in CEPT/GSM/WP2 [Working Party 2 in CEPT’s Groupe Spécial Mobile, with a mixture of French and English terminology] as technical expert during 1987. The role of CEPT changed later on, and the expert involvement disappeared, but at that time I was a member of Östen’s team,” Eriksson says.
POLITICAL INTRIGUES IN MADEIRA
But if the results of the Paris trials put an end to the technological GSM contest, the political rivalry continued. And there was a great deal of prestige at stake. The showdown came at the GSM meeting in Funchal on the island of Madeira from February 16–20, 1987.
Haug recalls: “This meeting was different from all the previous GSM meetings because the political element was so strong, an area where we engineers had no power. The fact there were no politicians there did not make things any less complicated.”
The Madeira meeting has acquired legendary dimensions in the history of GSM. The participants debated the pros and cons of wideband and narrowband TDMA into the small hours of the morning. Haug describes how starving delegates raided the conference center refrigerator after midnight one night. All they could find were tins of sardines, a whole stock of them. “They all went. And if our hosts ever noticed their loss, they did not say anything about it.”
Thirteen of the 15 countries represented at the meeting advocated narrowband TDMA according to the Nordic concept, and pointed to the trials. But France and Germany refused to abandon wideband TDMA, promoting the wideband solution proposed by Alcatel/SEL.
“At one point it seemed the outcome would be to choose both systems,” says Haug. “That would have had the absurd effect of dividing Europe into a Franco-German section incapable of communicating with the other countries. The engineers from France and Germany effectively admitted that our solution was better but they made it clear that if they went home and said that the Nordics had won, they would lose their jobs.”
Author: Svenolof Karlsson & Anders Lugn