[<< Back to Training Days overview]

Abstract

To allow for dynamic spectrum access, sensing techniques are crucial. Imec’s Sensing Engine can add such sensing techniques to radio systems and enables the evaluation of cognitive network solutions. The Sensing Engine consists of a flexible analog frontend or RFIC, combined with a digital frontend with sensing capabilities. Currently, there are two types of Sensing Engines integrated in the CREW infrastructure, provided by iMinds: one with imec’s analog frontend Scaldio2b coupled to imec’s digital frontend DIFFS, and one with the WARP analog frontend coupled to DIFFS. Both Sensing Engines can perform different modes of sensing. During the training, the Sensing Engine will be presented to the audience with a detailed description of its capabilities. A demonstration will be given, after which a hands-on tutorial will conclude the training.

Description

As the number of wireless devices and services for mobile communication increases, the radio spectrum is becoming more and more scarce. In addition, future communication systems are targeting heterogeneous co-existing networks, nodes and user devices. At the same time, energy efficiency is gaining importance in both devices and networks. To allow for dynamic spectrum access, sensing techniques are identified as a crucial component. The imec Spectrum Sensing Engine adds advanced sensing capabilities to radio systems and  enables, combined with the available CREW infrastructure, the evaluation of cognitive network solutions.

At imec, the CSI department (Circuits and Systems for ICT, formerly Green Radio) is developing hardware and algorithms for Reconfigurable Radio in mobile devices. Closest to the antenna there is the RFIC (Radio Frequency Integrated Circuit) or analog frontend, which takes care of the downconversion to baseband signals and analog-to-digital conversion (ADC). Next in the path there is the digital frontend for handling (amongst others) Automatic Gain Control (AGC), digital filtering and synchronization. To save power the next component (the baseband processor) is woken up only once synchronization has been found. Since both the analog and digital frontend are “on” most of the time, they should be low power, but it also makes them suitable for low power sensing. Therefore, both the Scaldio2b analog frontend as the DIFFS digital frontend are developed by imec targeting low power, low area and low cost to enable the use of these components in Reconfigurable Radio elements. These two components combine to form the imec Spectrum Sensing Engine.

Figure 1: schematic overview of imec’s Sensing Engine in a Reconfigurable Radio platform

Currently, imec has developed Sensing Engines in two different configurations. One with the in-house developed Scaldio2b analog frontend, and one with the commercially available WARP frontend, developed by Rice University. Both Sensing Engines are versatile in the way that they can easily be reconfigured for different standards and to perform different modes of sensing:

• FFT-based energy detection: on the DIFFS, the FFT accelerator calculates an FFT of the incoming signal, providing high resolution energy information
• fast energy detection: the DIFFS calculates the energy present in a certain part of the spectrum for a fast assessment of channel usage
• feature detection: the DIFFS performs hardware accelerated feature detection to detect the presence of a certain standard in the spectrum, e.g. DVB-T cyclostationary detection.

As one can see, the sensing algorithms are executed in hardware, on the DIFFS chip. This enables faster processing than calculations in software, which gives imec’s Sensing Engine an advantage over other sensing solutions.

Figure 2: left: imec Sensing Engine with Scaldio2b analog frontend; right: imec Sensing Engine with WARP analog frontend

Both Sensing Engine configurations are integrated in the iMinds wIlab2 testbed, made available by the CREW project. The iMinds testbed allows the user to experiment with the Sensing Engine without having to master the specific configuration of the Sensing Engine itself. Through the OMF interface foreseen in the iMinds testbed, sensing activities can be initiated together with configuration of the testbed. The results are stored in a database and can be accessed via straightforward SQL queries.

Content of the training

• the purpose of imec’s Sensing Engine
• overview of the different components
• the different modes of operation: specifications and generated output
• examples of usage
• hands-on: using imec’s Sensing Engine in the iMinds w-Ilab.t / wIlab2 - included in the iMinds hands-on on Wednesday

[<< Back to Training Days overview]