Cognitive solutions using imec’s spectrum sensing pave the way towards cognitive radios

Since the introduction of dynamic spectrum access and cognitive radio systems focus has been on techniques that allow capturing information from the environment. To enable distributed sensing and eventually map the spectrum environment, future radio platforms should ideally enable flexible spectrum sensing according to different scenarios, bands and algorithms.

An important target is the demonstration of a reconfigurable engine for multi-purpose spectrum sensing within the power and cost constraints of mobile devices.

This research takes place in the context of imec's research on cognitive reconfigurable radio systems (see section Reconfigurable radio baseband), (see section Reconfigurable radio front-end).

Reconfigurable sensing engine

The proposed reconfigurable sensing engine builds on a flexible analog front-end and a digital front-end for sensing (see fig. 1). It can be connected further to a baseband processor, resulting in sensing capable wireless devices. We further detail the functionality of the engine and its implementation.

Figure 1

Figure 1: Integrated sensing engine.

Flexible analog front-end

The reconfigurable analog front-end allows scanning the spectrum in different frequency bands. Since the basic task of a spectrum sensing engine is to scan all frequency bands for the presence of signals with several possible bandwidths, the functionality of the analog front-end actually requires the use of a software-defined radio (SDR). The flexible analog front-end is implemented with the 'scalable radio' SCALDIO radio frequency (RF)IC. SCALDIO is a fully reconfigurable analog transceiver implemented in 40nm CMOS technology. It covers all functionality from the low-noise amplifier (LNA) to the analog-to-digital converter (ADC), in a programmable way. The receiver RF operating frequency is programmable from 0.1 to 6GHz and the channel bandwidth is programmable between 1 and 40MHz. The high-speed network-on-chip used for configuration and the fast settling time of the phase-locked loop (PLL) enable fast switching between different RF frequencies and channel bandwidths. The low noise figure, 2.4 to 4dB below 3GHz, together with low power consumption, 30 to 90mA, makes this reconfigurable analog front-end very well suited for low-power flexible sensing.

Figure 2

Figure 2: SCALDIO flexible analog front-end architecture.

Digital front-end for sensing

The programmable digital front-end features the processing capabilities needed to analyze the signals in the scanned band.

  • being the interface between the analog front-end and the baseband (front-end data and control interfaces);
  • performing the signal acquisition and coarse time synchronization for the targeted standards;
  • multiple channelization branches, for simultaneous sensing and reception of different frequencies;
  • support for multi-band energy detection and feature-based sensing, relying on autocorrelation.

The sensing digital front-end is implemented with the digital front-end for sensing (DIFFS) architecture. DIFFS has now been fully implemented as a chip in 65nm CMOS technology. This will enable will enable hardware integration with the analog part and full real-time validation. The architecture consists of a fixed and flexible filter branch, which supports flexible band selection and resampling. The Sync/Sense engine consists of a single instruction multiple data (SIMD) processor with scalar slot, combined with several accelerator cores for synchronization and sensing.

Figure 3

Figure 3: DIFFS digital front-end for sensing architecture.

Demonstration setup

The demonstration setup consists of two printed circuit board (PCBs), one containing the analog front-end, one containing the digital sensing engine. Such a setup can be demonstrated to perform a frequency sweep from 500MHz up to 2.5GHz in only 7.6ms. Focus for demonstration is also on the sensing of long-term evolution (LTE) and/or digital video broadcasting-terrestrial (DVB-T) (two detection algorithms will be executed and compared: the first one is a low complexity energy detection algorithm; the second one is a feature-specific cyclostationairy detector).


Figure 4

Figure 4: SCALDIO hardware board.

Figure 5

Figure 5: DIFFS hardware board.

Figure 6

Figure 6: Integrated sensing engine.