Biomedical signal processing has become a very active domain of research nowadays. With the advent of portable monitoring devices, from accelerometer-enabled bracelets and smart-phones to more advanced vital sign tracking body area networks, this field has been receiving unprecedented attention. Indeed, portable health monitoring can help uncover the underlying dynamics of human health in a way that has not been possible before. Several challenges have emerged however, as these devices present key differences in terms of signal acquisition and processing in comparison with conventional methods. Hardware constraints such as processing power and limited battery capacity make most established techniques unsuitable and therefore, the need for low-complexity yet robust signal processing methods has appeared. Another issue that needs to be addressed is the quality of the signals captured by these devices. Unlike in clinical scenarios, in portable health monitoring subjects are constantly performing their daily activities. Moreover, signals maybe captured from unconventional locations and subsequently, be prone to perturbations. In order to obtain reliable measures from these monitoring devices, one needs to acquire dependable signal quality measures, to avoid false alarms. Indeed, hardware limitations and low-quality signals can greatly influence the performance of portable monitoring devices. Nevertheless, most devices offer simultaneous acquisition of multiple physiological parameters, such as electrocardiogram (ECG) and photoplethysmogram (PPG). Through multi-modal signal processing the overall performance can be improved, for instance by deriving parameters such as heart rate estimation from the most reliable and uncontaminated source. This thesis is therefore, dedicated to propose novel low-complexity biomedical processing techniques for real-time/online applications. Throughout this dissertation, several bio-signals such as the ECG, PPG, and electroencephalogram (EEG) are investigated. %There is an emphasis on ECG processing techniques, as most of the bio-signals recorded today reflect information about the heart. The main contribution of this dissertation consists in two signal processing techniques: 1) a novel ECG QRS-complex detection and delineation technique, and 2) a short-term event extraction technique for biomedical signals. The former is based on a processing technique called mathematical morphology (MM), and adaptively uses subject QRS-complex amplitude- and morphological attributes for a robust detection and delineation. This method is generalized to intra-cardiac electrograms for atrial activation detection during atrial fibrillation. The second method, called the Relative-Energy algorithm, uses short- and long-term signal energies to highlight events of interest and discard unwanted activities. Collectively, the results obtained by these methods suggest that while presenting low-computational costs, they can efficiently and robustly extract biomedical events of interest. Using the relative energy algorithm, a continuous non-binary ECG signal quality index is presented. The ECG quality is determined by creating a cleaned-up version of the input ECG and calculating the correlation coefficient between the cleaned-up and the original ECG. The proposed quality index is fast and can be implemented online, making it suitable for portable monitoring scenarios.