MIMO for a Mass-Market

Shifting the MIMO paradigm towards robustness and low cost

The mass-market for communication electronics requires cheap and energy-efficient solutions with a high relibility, taking into account increasing mutual interference. These solutions are expected to be found in effectively combining the technical possibilities in the RF and BB part of the radio receiver, and the system and implementation work. The design paradigm of products in a mass market seems to shift to (1) adaptive repression of interference, in which (2) adaptivity is partly realised in the RF frontend. Goal of the project is to gain design knowledge (FR/BB co-design) of wireless communication systems, with a focus on adaptive RF and adaptive algorithms for repression of interference. One is especially interested in MIMO (multiple input multiple output) technology, e.g., the use of multiple antennas on the sending, as well as on the receiving side. The intention is to design and build a flexible demonstrator.

Goal

Creating wireless communication for the mass market which is low cost and robust against interference and besides uses little energy through a good combination of analogue and digital signal processing.

Objectives

• Can we design, realize and validate a MIMO demonstration setup which conclusively combines strong robustness against interference with possibilities for electricity consumption and cost reduction?
• Can we reduce the costs and the electricity consumption of MIMO – RF transmitters and receivers by optimizing beyond the limit of RF and base band?
• Can we design a flexible base band platform which is fitted to MIMO technology and which we can use to prove the potential for cost and electricity reduction?
• Can we improve the robustness of MIMO systems against low costs?

Combination of analog and digital signal processing

Further increase in the use of WLAN systems offers great challenges in the use of the scarce RF spectrum, the occurring interference, the electricity consumption in mobile applications and the cost price. Worldwide research is being conducted into systems for wireless communication systems within homes, among which MIMO. These projects aim particularly at high speed. Searching for a solution which is low cost, robust against interference and which uses little energy because of a good combination of analogue and digital signal processing is unique.

Results

We considered a WLAN scenario, where a basestation receives the desired user signal, along with interfering signals from neighboring users. Using multiple antennes (MIMO), it is possible to do source separation and suppress the interfering signals via beamforming. Typically, this is done in digital baseband. This implies that all antenna signals are digitized and available in baseband. As Analog to Digital Converters (ADCs) are expensive and power-costly, we have studied the possibility of doing beamforming in the analog domain, thus reducing the number of channels before the ADCs, leading to direct power savings. Further, indirect power savings are possible by interference suppression via the (analog) beamforming, as the required dynamic range of the ADCs will become significantly smaller (reduced headroom).

Using this and similar architectures, the results are as follows:

• For narrowband cases, beamforming is sufficient and further equalization is not necessary. In this case, we proposed a design for the beamformer coefficients in the analog domain (the Analog Preprocessing Network or APN), and the subsequent beamforming in digital domain that leads to the desired output signal. The design is based on the number of ADC units and the number of bits used by each ADC, as this directly determines the power consumption, in relation to the reached performance (Mean Squared Error at the output). Further, we realized via discussion with the other project partners that in practice APNs are poorly quantized. We offered a design for finding the optimal beamforming vectors in this case.
• For wideband cases, which would be the case in practical systems, the situation is more complex, and we don’t have complete solutions. However, we do have suboptimal but practically meaningful extensions.
• In another direction, we study an architecture where we subtract from the input of the ADC units the estimated contribution of the interferers, thus reducing the dynamic range and power consumption. The interferers are estimated from their residual components at the output of the quantization. The resulting feedback structure ("feedback baseband beamformer") is nicely combined with currently widely used Sigma-Delta ADCs, that also have a feedback structure. We offer a closed-form design of the beamforming coefficients, based on a training sequence for the desired user. This is a revolutionary design; its full implications need further study.
• Finally, at the start of the research, we have studied superimposed training sequences in the context of unsynchronized OFDM. Current WLAN signals are all based on OFDM modulations; and neighboring interferers will not be synchronized. In general, it is very hard to develop MIMO receiver algorithms for this case. We showed that, by inserting training sequences on top of existing WLAN signals, it is possible to zoom in on only the desired signal, and suppress the interfering signals. For narrowband cases, this works very well, as we had demonstrated before when we derived the KMA algorithm. For wideband cases (as would be the case in practice), space-time equalizers are needed; the KMA cannot be applied, and new algorithms had to be derived.

Start:	Sept. 2005
End:	Sept. 2009
Sponsor:	IOP Gencom (Senter-Novem)

Partners:

• TU Eindhoven (Jan Bergmans, Frans Willems, Peter Baltus)
• Univ. Twente (Cees Slump)
• Industrial partner: Philips Research (Jean-Paul Linnartz)

Links

• Publication list (CAS part)
• People
• Project leaflet

Contact address

Mail:	prof.dr.ir. Alle-Jan van der Veen
	Delft University of Technology
	Fac. EWI/Electrical Engineering
	Mekelweg 4
	2628 CD Delft
	The Netherlands
Phone:	(+31 15) 2786240
Fax:	(+31 15) 2786190
E-mail:	allejan@cas.et.tudelft.nl