Dresden is like Silicon Valley: hungry talent & excellent education

Fettweis: The world is connecting to the Internet, and users do it increasingly by means of wireless technology. What WiFi has been and still is for the fixed-line user, 3G and 4G will be for wireless users. People today use UMTS and HSPA much as a data modem for their mobile PCs. For future applications users will need data rates associated with LTE. Only then will they be able to receive, for instance, real-time video streaming in acceptable quality, or, as an SME, connect to the Internet and operate in international markets. Wireless broadband is a prerequisite for the next round of economic globalization. To connect the world wirelessly to the Internet is one of the most important infrastructure challenges for the future. A second driver is sensor technology. Going forward, hundreds of sensors per person will need to be connected. In order to enable this change, mobile telecommunications systems have to be adapted to this task.

EE Times: What does this mean?

Fettweis: Because the protocols required are currently not designed for this purpose. They have their technical roots in telephony; all the systems today are designed for voice with very specific requirements. We have to get energy consumption down by a factor of 1000. And I believe this is feasible – for the sensor end devices. We will see terminal devices 1000 times more energy efficient than today.

These are the two main challenges: Ubiquitous [wireless] broadband and wireless sensor networks. At the recent ICC 2009 [the IEEE International Conference on Communications, held in Dresden] we had an LTE advanced workshop, we had a lot of sensor network presentations, we had next generation Internet with focus on protocols and infrastructure. And there was great focus on energy efficiency, the consensus being that if we want Internet connection everywhere, we need much higher energy efficiency.

Today about 10 percent of the global power generation is used by the information and communication technology infrastructure. If this continues by a factor of 2 every five years, as was the case over the past years, ICT will use 40 percent of the power generation capacity by 2020. This is unsustainable and means that we have to bring down power consumption for ICT infrastructure including mobile communications by one magnitude in order keep power consumption at today’s percentage level.

For these reasons, we had a ‘green communications’ workshop at ICC 2009. Power efficiency also has been a topic of my own research for some time, and the ‘Cool Silicon’ research and industry cluster here in Dresden is clearly going into this direction.

EE Times: One of the topics of your chair at TU Dresden is implementing wireless architectures. How can your scenario be transferred to the chip level?

Fettweis: We are focusing on high data rates for LTE and LTE Advanced architectures and energy efficiency at the same time. We have built our ‘Tomahawk’ demonstrator with almost 60 million transistors which proves that it is possible to get the power consumption down by a factor of ten compared to today’s low energy concepts provided by Qualcomm, Ericsson and the like.

EE Times: Are available chip design tools adequate to do such complex design work? Do these design projects trigger activities to further develop EDA tools?

Fettweis: Yes, massively. We had to develop new design methodologies in the course of the Tomahawk chip project. The Tomahawk integrates 14 processors – two from Tensilica and twelve designed by ourselves, but generated by an EDA framework. Despite this, they are very energy efficient. These are the reasons why we discuss these things with, for instance, Synopsys and CoWare. Currently we are in talks with Tensilica to get their tools. The EDA industry is very interested in what we do here.