Volume 66, Issue 9-10 -

Software defined radio (SDR) clouds centralize the computing resources of base stations. The computing resource pool is shared between radio operators and dynamically loads and unloads digital signal processing chains for providing wireless communications services on demand. Each new user session request particularly requires the allocation of computing resources for executing the corresponding SDR transceivers. The huge amount of computing resources of SDR cloud data centers and the numerous session requests at certain hours of a day require an efficient computing resource management. We propose a hierarchical approach, where the data center is divided in clusters that are managed in a distributed way. This paper presents a set of computing resource management tools for analyzing computing resource management strategies and algorithms for SDR clouds. We use the tools for evaluating a different strategies and algorithms. The results show that more sophisticated algorithms can achieve higher resource occupations and that a tradeoff exists between cluster size and algorithm complexity.

Software-Defined radio (SDR) is a future-proof solution for designing flexible and adaptable wireless networks and equipments. It replaces conventional radio hardware with reconfigurable, reprogrammable radios. A graphical approach for designing flexible SDR multi-standard systems is proposed, which provides all the possible alternatives of implementation capable of realizing the multi-standard design. However, a cost function which evaluates the cost of any one of these options is proposed in previous work. All these ideas are briefly mentioned in this paper but however, our goal is to help finding the option of implementation which has the minimum cost. Graph theory is adopted and particularly the study of directed hypergraphs, to present a new idea algorithm capable of solving this optimization problem. This algorithm provides an exact-optimal solution, unlike the previously applied heuristic methods which give a near-optimal solution. Furthermore in this work, we analyze the computational complexity of our algorithm.

In this article, we propose a flexible CSMA based MAC protocol which facilitates research and experimentation using software define radios. The modular architecture allows to employ the protocol on platforms with heterogeneous hardware capabilities and provides the freedom to exchange or adapt the spectrum sensing mechanism without modifying the MAC protocol internals. We discuss the architecture of the protocol and provide structural details of its main components. Furthermore, we present throughput measurements that have been obtained on an example system using host-based spectrum sensing.

This article is about a Spectrum Sensing Network (SSN) which generates an accurate radio environment map (e.g. power over frequency, time, and location) from a given application area. It is intended to be used in combination with cognitive Program Making and Special Events (PMSE) devices (e.g. wireless microphones) to improve their operation reliability. The SSN consists of a distributed network of multiple scanning radio receivers and a central data management and storage unit. The parts of the SSN are presented in detail and the advantages and use cases of such a sensing network structure will be outlined.

Based on the sparse property of the Cyclic Autocorrelation Function (CAF) in the cyclic frequency domain, we propose in this paper a new blind estimator using compressed sensing, which better estimates the Cyclic Autocorrelation Function at a given delay compared to the classical non blind estimator when using the same number of samples. Two metrics are used to evaluate this estimation: the Mean Square Error of type one and that of type two ( MSE 1 and MSE 2 ). The first metric compares the estimated Cyclic Autocorrelation Vector (CAV) which is a particular vector of the CAF for a fixed delay τ , with the theoretical reference vector obtained using the CAF. The second metric compares the value of the estimated cyclic frequency with the theoretical one. Simulation results show lower ( MSE 1 and MSE 2 for the new proposed estimator compared to the classical one under the same conditions, from the simplest scenario to the more realistic one that uses a filter at the transmission side, a reception noise, and fading channel.

Cyclostationary signatures have been shown to be an effective method for OFDM network synchronization and Cognitive Radio coordination. In this article, an extended method that utilizes cyclostationary signatures for signal parameter identification of OFDM-based Cognitive Radio nodes is presented. The scenario, implemented on a GNU Radio based evaluation platform, shows how different signal parameters, e.g. carrier frequency, occupied bandwidth and the used modulation scheme can be identified at the receiver side using the described approach. A major drawback of cyclostationary detection in OFDM systems is its sensitivity to frequency offset and sampling rate mismatches between oscillators at the transmitter and the receiver. An analytical model that characterizes this impairments is derived, followed by a discussion of implementation issues and the performance evaluation of proposed cyclostationary signature detection, both in a simulation environment and through RF experiments.

Software defined radios are increasingly used in modern communication systems, especially in cognitive radio. Since this technology has been commercially available, more and more practical deployments are emerging and its challenges and realistic limitations are being revealed. One of the main problems is the RF performance of the front-end over a wide bandwidth. This paper presents an analysis and mitigation of RF impairments in wideband front-ends for software defined radios, focussing on non-linear distortions in the receiver. We discuss the effects of non-linear distortions upon spectrum sensing in cognitive radio and analyse the performance of a typical wideband software-defined receiver. Digital signal processing techniques are used to alleviate non-linear distortions in the baseband signal. A feed-forward mitigation algorithm with an adaptive filter is implemented and applied to real measurement data. The results obtained show that distortions can be suppressed significantly and thus increasing the reliability of spectrum sensing.

Over the last years ongoing advances in ADC technology have enabled RF signals to be sampled at IF frequencies. Undersampling is nowadays employed in software-defined radio or radar receivers and offers the possibility to relieve requirements in the analog receiver partition. Unfortunately, when moving to higher IF concepts, this becomes demanding for the ADC itself, because of inherent spurious-free dynamic range (SFDR) roll-off that increases with input frequency. This fact often limits the receiver's IF placement to Nyquist zone (NZ) 2. In this work the emerging concept of Digital Assistance is pursued to give the receiver access to higher NZs while making no compromise on the SFDR. We will present and discuss post-correction results for two 16-bit high-speed converters from two different vendors at 120 and 125 MSPS, respectively. The proposed system-level post-correction decomposes nonlinearity into a static and a dynamic part. For both ADCs under investigation the degraded SFDR in higher NZs could be improved by up to 15 dB using purely digital linearization technologies, thus increasing the detectability of small signals in the presence of very strong signals or interferers. Near-identical results for both ADCs confirm the general validity of the system-level correction approach.