Home // SnT // Research // SIGCOM // Communications Lab

SIGCOM CommLab

Contact: Juan Merlano DuncanStefano Andrenacci and Symeon Chatzinotas

The Interdisciplinary Centre for Security, Reliability and Trust’s Signal Processing and Communications Research Group has developed the SIGCOM CommLab (Communication Laboratory) as a natural consequence of its intensive research activities on signal processing techniques for both terrestrial and satellite communication systems.

The CommLab is mostly focused on wireless communication systems for testing and validating digital signal processing algorithms while facing real implementation issues and constraints on real-time hardware platforms. The aim is to develop end-to-end/Proof of Concept through real-time wireless communication testbeds which are used to model, design and test the digital signal processing algorithms studied by the SIGCOM group for satellite/terrestrial systems. The applications under development vary from interference management techniques for satellite communication, emulation of satellite communication systems, spectrum sensing for Cognitive Radio (CR) technologies, Spectrum Monitoring and Multiple-Input Multiple-Output (MIMO) systems.

Figure 1: Picture of the SIGCOM CommLab

The general infrastructure includes a range of SDR development platforms by National Instruments, each of them connected to a central hub used for selecting the sub-infrastructure required for the specific test while also providing control and monitoring functionalities. Each board is itself a single-antenna/multi-antenna system equipped with a radiofrequency module, digital-to-analog and analog-to-digital converters and a high performance FPGA for digital processing. The central hub is also supported by a workstation with high computational capabilities, equipped with a powerful and programmable GP-GPU by NVidia, which is used to speed up all the off-line calibration procedures required by the algorithms.

The main advantages of the infrastructure arise from its capability as a modular, reconfigurable and scalable system which can be controlled and monitored by the same central control unit, allowing for flexibilities in implementing and testing a plethora of communication scenarios.

Figure 2: Pictorial Scheme of the Infrastructure.

Figure 2 presents a pictorial scheme of the overall infrastructure.

Included in the infrastructure, four 2x2 USRP-RIO are connected to the control center (PXI System) through a PCIe interface (MXI interf) which leads to the possibility of testing up to an 8x8 MIMO system. Thanks to its modularity, the system is also easily expandable to high order MIMO systems since the PXI can count on additional 7 free-slots. In order to support all the computational load needed by such a MIMO systems, 2 FPGA boards are foreseen in the infrastructure. All the mentioned boards, both the USRP-RIOs and the FPGAs, are connected by a dense network of PCIe busses which make the interconnections both high throughput and low latency.

The synchronization of the USRPs, especially in the transmission part, is guaranteed in the infrastructure by a dedicated board (Accurate Clk) able to generate a very accurate reference signal to be provided to all the boards. This board is directly connected to the clock distributor (Octoclock-G) with the aim of splitting the same reference clock to all the boards.

For measurement purposes, a vector signal transceiver (VST) is included in the system ad it is usually used as a real-time Spectrum Analyzer.

The control unit (host) is also comprised in the system. It is basically a workstation which is directly connected to the overall system and this of course provides some advantages basically in terms of throughput that the host can manage.

The host control unit is connected by means of Ethernet connection to an external workstation equipped with a general purpose GPU which can be used to speed up the off-line software to be pre-computed.

The workstation directly controls four USRP N200 through an Ethernet connection which can be also used either for MIMO systems or for single transceivers.

SERENADE: Satellite Precoding Hardware Demonstrator

 
The SERENADE research project at SnT

SERENADE is a demonstrator of Full Frequency Reuse for Multibeam Satellite Systems using Multi-User Multiple Input Multiple Output (MU-MIMO, also known as Precoding). This includes the Symbol Level Precoding algorithm, which optimizes the precoding process in real time for every transmitted symbol. This technique exploits the constructive interference among multiuser links and gives a phenomenal improvement in throughput, availability and energy efficiency. (read more)

Figure 3: Satellite Multi-beam Scenario

Frame Based Precoding. Objective: to demonstrate the gains of frame-based precoding in comparison to conventional precoding techniques.

Figure 4: Frame-Based Precoding.

[REF 1] D. Christopoulos, S. Chatzinotas, and B. Ottersten, “Frame based precoding in multiuser multibeam satellite systems," IEEE Trans. Wireless Commun., 2014, (under review), preprint: arXiv:1406.6852 [cs.IT].

[REF 2] D. Christopoulos, S. Chatzinotas, and B. Ottersten, “Multicast Multigroup Precoding and User Scheduling for Frame-Based Satellite Communications," IEEE Trans. Wireless Commun., 2015, (under review), preprint: arXiv:1406.6852 [cs.IT].

Symbol level Precoding. Objective: to demonstrate the gains of symbol-level precoding in comparison to conventional precoding techniques. 

Figure 5: Symbol level Precoding.

[REF 3] M. Alodeh, S. Chatzinotas and B. Ottersten, “Constructive Multiuser Interference in Symbol Level Precoding for the MISO Downlink Channel,” accepted in IEEE Trans. Sig. Proc., Available on arXiv:1401.4700 [cs.IT].

[REF 4] M. Alodeh, S. Chatzinotas and B. Ottersten, “Energy-Efficient Symbol Level Precoding in Multiuser MISO Based on Relaxed Detection Region,” submitted to IEEE Trans. Wir. Comm., 2015.

The test-bed includes the generation of different streams for different users, the transmitter part, the channel and the receiver module. In Figure 5, a general scheme of the test-bed for the forward link of a satellite multibeam link between the GW and user terminals is shown. The test-bed is able to generate 8 different streams, which are then coded and modulated according to DVB-S2X standard for FWD link transmission (plus some other internal private air interface structures), to emulate the feeder link and the multibeam satellite for the transmission of the precoded waveforms and a programmable set of receivers able to calculate the Channel State Information to be fed-back to the GW for Precoding purposes.

Figure 6: General block scheme of the satellite test-bed configuration.

The test-bed has been presented at the SES iD2018 days, one of the most important events for the SatCom community, especially for industries operating in the SatCom market.

Figure 7: Satellite Precoding Demonstrator at the SES iD2018

Satellite Multi-Channel Emulator

The channel emulator encompasses the effect of the whole satellite forward link, from the IF input at the RF equipment in the gateway to the IF output of the LNA/LNB in the terminal user equipment. This makes a transition from a laboratory demonstrator to real live satellite testing less problematic.

It is able to emulate satellite channels for:

  • GEO and Non-GEO satellite links
  • MIMO, MISO, SIMO and SISO channels

For each channel chain, it generates:

  • IMUX/OMUX filters, given taps as input. One of the models used is the one proposed in the DVB-S2 standard (ETSI EN 307 V1.2.1). Other proprietary non-disclosed models provided by the project partners are used.
  • TWTA characteristics, given AMAM, AMPM characteristics as input. These inputs are stored in Look-Up-Tables inside the FPGA. These parameters can be updated at any moment and even during runtime. The linearized and non-linearized models from the DVB-S2 standard are default options.
  • MIMO downlink (user link) channel: starting from a given antenna pattern, by selecting the desired user positions, a channel matrix is generated and loaded by the channel emulator, which then applies the matrix using the whole available transmitted streams
  • RF impairments (AWGN, Phase Noise, frequency errors,doppler): The channel emulator has a high speed AWGN implemented in the FPGA with a pseudo-random number generator developed by the SnT group. The power of the noise can be accurately adjusted to have a desired power level. This generator models the noise produced at the user terminal but also takes into account the aggregated noise over the whole forward link. The frequency offset is easily generated with a slight change of the carrier frequency of the USRP platform. The aggregated phase noise of the complete forward link (mainly affected by the user-terminal phase noise) is included in the channel emulator inside the FPGA, and used a set of PRNGs. The phase noise mask can be modified at any time.
  • Satellite channel impairments (e.g. rain fading)

In summary, all the payload emulator parameters can be updated during run time. This feature gives the channel emulator the capability to introduce all the required payload imperfections for different scenarios with complete flexibility. The payload parameters can be loaded from an external files, or can be generated locally from desired functions.

It is worth noting that the use of a satellite channel emulator which implements not only the payload and receiver impairments, but also the linear combinations of the different carriers through a channel matrix, is fundamental for a proper execution of the in-lab testing and for the performance characterisation, especially in MIMO channels and in experiments which involve precoding techniques.

The channel emulator can also be announced re-configured in order to emulate multi-channel scenarios as, for example, in Non GEO links where the link budget and the impairments are user terminal specific. In fact, considering a LEO satellite and two user terminals placed in different positions over the coverage, the two channels, at a specific time stamp, show different signal to noise ratios, doppler effects and delays.

Spectrum Sensing using USRPs

Spectrum utilization can be significantly improved by adopting cognitive radio (CR) technology. Such radios are able to sense the spectral environment and use this information to opportunistically provide wireless links that meet the user communications requirements optimally. To achieve the goal of cognitive radio, it is a fundamental requirement that the cognitive user (CU) performs spectrum sensing to detect the presence of the primary user (PU) signal before a spectrum is accessed as to avoid harmful interference.

Several spectrum sensing techniques have been proposed in the literature, which with some modifications can be extended in the detection of interference, such as matched filter [3], energy detector, feature detectors and eigenvalue detectors.

Among these methods, the ED is the most popular one because of its simplicity. However, it needs accurate knowledge of signal/noise variance (in the case where interference has to be detected), thus is very sensitive to the phenomenon of variance uncertainty.

The setup is a scenario where a single antenna is employed by the receiver to detect interference, while the transmitter and the interferer have also only one transmit antenna. This way, the detection problem can be formulated as the following binary hypothesis test:

H : x[n] = s[n] + w[n] n = 0, .., N − 1

H1 : x[n] = s[n] + w[n] + i[n] n = 0, .., N – 1

where w(n) is the additive noise at the receiving antenna, modelled as an independent and identically distributed (i.i.d.) complex Gaussian vector with zero mean and variance σ2 w , s(n) is the desired signal at the receiving antenna and i(n) is the interfering signal at the receiving antenna to be detected.

The above scenario can be tested on Software Defined Radios using USRPs communication platform which can be interfaced and programmed by model-based MATLAB/Simulink environment.

 

 

Figure 6: Experimental Setup.

 

In Figure 6 a picture of the experimental setup is shown, where as mentioned earlier, our communication system consists of three USRP N200, two of them are used as the transmitter of the desired and interfering signal, respectively, and the third one is employed as the receiver. Furthermore, Figure 7 shows the block diagrams of the transmitter and receiver.

 

 

Figure 7: Block diagrams of the transmitters and the receiver.

 

Moreover, Figure 8a presents the power spectral density (PSD) of the received signal (bandwidth of 5 MHz) without the presence of interference. In this case, only two USRPs are power on, the desired transmitter and the receiver. The energy of the received signal is computed in order to establish the amount of the energy, where we are supposed to get. This quantity is stored in the memory of the receiver and then, the third USRP (interfering transmitter) is activated, generating a narrowband signal with bandwidth of 1.25 MHz and the PSD is shown in Figure 8b. In this second case, it is obvious that the presence of interference increases the energy of the received signal, thus, selecting a proper threshold, the ED can be a good detection scheme. Finally, Figure 9 illustrates the probability of interference detection versus the ISNR for a fixed PFA =0.1.

 

     

Figure 8: a) PSD of the received signal without interference. b) PSD of the received signal with narrowband interference.

 

Figure 9: Probability of interference detection for a fixed PFA =0.1 and number of samples, N=25.

 

 

Power and Direction of Transmission Estimation (DoT)

In this specific scenario, the spectrum monitoring, the goal is to estimate the transmission power and direction of transmission of a directive source as in Figure 10.

Each USRP receives the observation samples, and sends the raw data to the main hub which can be a PC or a DSP. The DSP infers the incoming observations to determine the power and DoT. In SnT, we already have the required algorithms to do so. In later stages, we plan to look into algorithms where part of the processing is done at the USRPs before sending the pre-processed data to the main hub. However, most of the complex processing algorithms shall be performed in the main hub.

 

Figure 10: System setup for power and DoT estimation.