Matlab Modeling of Zero-IF Radio Receivers by Using Object Oriented Programming Botond Sandor Kirei 1 , Marius Neag 1 , Tomina Fabiola Sălăjan 1 , Marina Dana Ţopa 1 1 Faculty of Electronics, Telecommunications and Information Technology, Basis of Electronics Department, Technical University of Cluj-Napoca E-mail: Marius.Neag@bel.utcluj.ro Abstract – This paper presents a novel approach to the model- ing of radio receivers, based on object-oriented programming in Matlab. The RF/analog blocks are represented by analyti- cal models, obtained by selecting, configuring and merging of pre-defined attributes; these attributes are closed-form analyt- ical expressions of the basic analog signal-processing func- tions (signal amplification and filtering, frequency conversion, analog-to-digital conversion, etc.) and the non-idealities con- sidered (noise, nonlinearity, carrier frequency offset, I/Q im- balance, etc.). This method enables the development of well- structured yet fully scalable and re-configurable models, which are easy to maintain and refine further, by adding new features. A framework for modeling OFDM receivers was developed using this method; it comprises OFDM signal gen- erators and accurate representations of the operations per- formed in digital baseband, from digital channel filtering to error correction and OFDM decoding. Also, several signal quality monitors such as the Signal to Undesired-Signal Ratio (SUSR) and Signal-to-noise and distortion ratio (SINAD) were implemented. The usefulness and flexibility of the proposed approach is demonstrated by an example, the modeling and analysis of a DVB-H tuner for both passband and equivalent baseband simulations. 1 Introduction Access to information and mobility are the key ingredients of building the knowledge-based society. These two ingredi- ents commission the electromagnetic (EM) spectrum to be the primary vehicle for the distribution of information. The con- tinuous development of wireless communications lead to new paradigms such as software defined radio (SDR) and cognitive radio (CR). SDR proposes the implementation of an universal user terminal (compatible with multiple standards) consisting of a “minimal” RF front-end followed by a high-performance ADC and an extensive digital signal processing (DSP) section which deals with channel selection, as well as demodulation, etc. CR brings wireless communication to the next level, by allowing the terminal to choose the band of operation. Auton- omous band selection became necessary because the licensed EM spectrum is close to saturation. A follow-up of the SDR and CR concepts is the “dirty RF” (DRF) paradigm, proposed in reference [1]. DRF encourages RF/analog block non-ideality assessment by the means of blind compensation algorithms. Reference [2] describes a digital compensation chain for the improvement of digitally- intensive radio receivers; the chain comprises three digital blocks inserted between the RF tuner (RF/analog front-end plus ADC) and the main DSP: a droop filter to compensate for the group delays caused by analog filters, an adaptive filter for reducing the effects of tuner non-linearity and a digital I/Q imbalance compensation. Obvious additions to this chain are the compensation of DC offset and carrier frequency offset (CFO); digital algorithms for reducing the effects of the fre- quency synthesizer phase noise can be considered, as well. In order to develop and verify such compensation algo- rithms one needs effective and accurate models for the RF tuner and a systematic way to generate test stimuli. The RF and analog blocks can be verified by running tran- sistor-level simulation, as described in [3]. This approach is limited by the tradeoff between simulation time and precision. Also, it is difficult to distinguish between the effects of vari- ous non-idealities, such as the noise and nonlinearity.. System level simulation can be conducted by using behav- ioral and analytical models. In reference [4] a method for developing behavioral models is proposed, based on Volterra series: first, a transistor-level analysis is performed on the RF/analog block being modeled, with the aim of “extracting” the nonlinearity features of the circuit; these features are then represented by specific coefficients of Volterra kernels. A simpler and more popular way to perform system-level analysis is based on analytical models [5], [6]. A complete receiver model that uses third parity IP to generate OFDM signals and comprises functional representations of DSP per- formed in the receiver can be found in [7]. The method pro- posed in this paper falls into analytical model category but employs a different approach: first, all the analog signal- processing functions (such as signal gaining, mixing, filtering, etc.) as well as the related non-idealities considered (such as noise, non-linearity, I/Q mismatch, quantization noise, etc.) are represented using closed-form analytical expressions; these representations will be called hereafter attributes. Next, mod- els for each RF/analog block are created by properly selecting, configuring and merging of the attributes; these models are called here composite models. Finally, a system-level model of the RF tuner is obtained by connecting the RF/analog block models according to the tuner architecture. A framework based on object oriented programming (OOP) was developed using this method that provides a systematic yet flexible ap- proach to the modeling of RF tuners. The paper is organized as follows: Section 2 details the ab- straction layers that had to be defined not only for the OOP implementation, but also for the better understanding of com- posite modeling. Section 3 briefly presents main features of OOP employed here. The proof of concept and simulation VOL. 2, NO. 4, DECEMBER 2011 72