In Sweden, ERA and Televerket were each trying during the early 1980s to decide which direction to take in digital mobile telephony. One of the main choices was between FDMA and TDMA, and to begin with they both focused on FDMA and then switched to TDMA.
Mäkitalo carried out his first FDMA experiment in 1977 in order to test whether NMT could be made to operate digitally. He identified two problems: that the narrow frequency channels slowed the data down and that speech quality was too poor. Six years later, he and his colleagues realized that it would be possible to create a digital system using FDMA, but by then had shifted their focus to TDMA. “TDMA offered greater future possibilities even though there were still problems to be solved.”
Uddenfeldt had the same experience: “We gradually came to realize that TDMA was the interesting technology. The FDMA trials were really important because they showed that it was possible to construct a digital system that was better than the existing analog ones. But TDMA would offer even more benefits. Several users could share the same radio equipment, which meant lower costs for building networks.”
In the spring of 1983 the development engineers at ERA and Televerket met in Televerket’s head office in Farsta. The meeting concluded with an agreement to focus on unresolved problems in TDMA. The main issues were propagation paths and intelligent equalization algorithms. From then on the partners met regularly to compare their findings.
During 1984 and 1985, ERA developed a system concept called DMS 90. This was based on several narrowband TDMA technologies that were later all included in the GSM standard.
Uddenfeldt gives some examples: “Interleaved channel coding combined with frequency hopping: this was a technology that ERA had used in military contexts and learned how to control. It provided considerably improved performance compared with analog systems.
“New handover techniques enabled the introduction of small cells only a few hundred meters in radius. The time gap between transmissions with TDMA meant the telephone could take measurements from the surrounding base stations. This was vital to allow us to raise the number of users in the network, which would otherwise have been swamped with signals.
“Advanced equalization techniques could cope with the radio reflections that occur during radio transmissions. These reflections can be delayed by a few microseconds, which means a TDMA system cannot work without equalization.”
This concept was described in internal ERA documents during 1985 and later tested during work on GSM. The equalization function posed the major challenge.
In the autumn of 1985, Håkan Eriksson, a new engineering graduate from Linköping, met Uddenfeldt for a job interview. He made a bigger impression than he realized. As they toured ERA’s laboratory, off the cuff he supplied a couple of algorithms needed for the initial work on the GSM.
These algorithms had been part of Eriksson’s degree project. He spent his final year of study at Stanford University, where he read everything he could find about digital radio technology. “But when I began at ERA, I knew nothing about GSM. On the other hand, I could see that what I had learned at Stanford was absolutely right,” Eriksson says.
Author: Svenolof Karlsson & Anders Lugn