Physical Layer Architecture

WP4 is working towards its set objectives and summary of technical highlights is as follows:

Transceiver Reference Model

The Transceiver block (TRX) needs to communicate not only with the Spectrum Sensing block (SS), but also with the Adaptation Layer (AL). Figure 1 shows the interfaces between the TRX, SS, AL and the CM-RM.

The red arrows represent either data or sensing paths. The black arrows represent control interfaces. The arrow indicates the logical direction of the information flow.

The TRX block includes all the functions required for the baseband processing of the data to be transmitted or received over the air, up to the conversion between analogue and digital domains. It is expected to be backward compatible to existing wireless standards (to be defined in detail in future work) and exhibit the flexibility, when demanded from a high-level entity like e.g. CM-RM, to evolve to a Cognitive Radio baseband processing block, with the ability to implement the particular functions of such a system.

The TRX block needs to interact with the SS block to gather information about the channel state. The TRX will receive the sensing control signals from the sensing block and it will send sensing measurements and PHY constraints to the SS block. Information regarding the frequency white spaces, the sensitivity of the channel and etc. are sent to the SS block to be compared with the sensing measurements and decide on the control signal. The interface between the AL and the TRX is two-fold. The TRX sends the data, physical layer metrics and constraints to the AL which then passes this to the higher MAC or networking layers. The AL in turn configures the TRX based on the inputs it receives from higher layer blocks and also from the CM-RM.

Flexible PHY for white spaces

Based on the scenarios and requirements, defined in WPs 1 and 2, Work Package 4 proposes alternative physical layer definitions for the white space devices. We put strong emphasis on the adjacent channel characteristics and spectrum pooling capabilities of the proposed waveforms. The WP investigates various modulation techniques for the slightly different requirements that stem from the diverse scenarios and applications. While proposing modifications to the conventional OFDM(A) schemes, retaining as much compatibility with legacy waveforms as possible, we also propose more general modulation schemes with better characteristics. The WP identifies the best modulation scheme for specific scenarios and proposes the corresponding adaptation of the transmission parameters. We also focus on the integration of the sensing and the payload transmission. The physical layer(s) defined in this WP will serve as an input for the design of the higher layers, and selected elements will also be used in the project demonstrations.

We investigate the use of windowed OFDM signals with fast out-of-band decay to meet the stringent spectral masks imposed on CR systems. We focus on the vestigial symmetry class of pulses, among which the raised cosine family stands out, allowing ISI free transmission. We also investigate the design of functions with vestigial symmetry and several null derivatives at the extreme points. These provide the required asymptotic decay in the transform domain, together with the use of higher cardinality constellations to compensate the spectral efficiency reduction imposed by the increased duration of the designed pulses. The WP also considers a specific combined time and frequency domain processing called IA-PFT which aims directly at improving the spectral characteristics of OFDM using cancellation carriers and time windowing separately.

Breaking the legacy of OFDM, the WP also considers Filter Bank Multicarrier (FBMC) schemes as strong candidates for the proposed white space physical layer. FBMC exhibits extremely low out-of-band leakage and the lack of the cyclic prefix results in better spectrum efficiency compared to OFDM. In addition, to enable efficient simulation of the complete cognitive radio system, an appropriate model of the proposed physical layer(s) must be given. The sensitivity of the proposed scheme(s) to real-world impairments and the computational complexity of possible implementations are also important for a viable implementation. The WP also examines the impact of some FBMC-specific challenges, i.e. channel estimation and equalization, adoption of various multi-antenna techniques etc.

Finally, a relatively new PHY design technique, Generalized Frequency Division Multiplexing (GFDM) was introduced. It has the flexibility of shaping the pulses used in the OFDM scheme so that these have lower out-of-band radiations to cause interference to the incumbent signals. However, the pulse shaping filter introduces inter carrier interference which degrades the performance of the GFDM system. We investigate how this degradation can be reduced by means of interference cancellation techniques. We introduce a basic serial and a double sided serial cancellation procedure suitable for GFDM and demonstrate that the self-introduced interference caused by the pulse shaping can be mitigated successfully.

Concept of an FBMC-based PHY for cognitive TVWS scenario

The PHY layer hardware concept is portrayed in Figure 2. It consists of several modules: the baseband board, the RF transmitter front-end, and up to two RF receiver front-ends. The digital signal processing as well as the hardware (re)configuration is performed by the baseband board, which includes an FPGA, an analogue-to-digital converter (ADC), and a digital-to-analogue converter (DAC). The analogue signal processing is performed by the front-end boards. This includes the frequency conversion, amplification, and filtering of the received signals and the signals to be transmitted, respectively.

A detailed description of the hardware concept as seen in figure 2 and the hardware implementation can be found in D7.2, as this is the basis for the 2^nd Proof-of-Concept demonstration. In the following sections, a general view on the hardware concept and the requirements are given that eventually led to the concept selected.