Adaptive Compression and Optimization for Real-Time Energy-Efficient Wireless EEG Monitoring Systems Ramy Hussein † , Amr Mohamed † , and Masoud Alghoniemy ∗ † Dept. of Computer Science and Engineering, Qatar University, Doha, Qatar. ∗ Dept. of Electrical Engineering, University of Alexandria, Egypt. {rhussein;amrm@qu.edu.qa, alghoniemy@alexu.edu.eg} Abstract—Recent technological advances in wireless body sen- sor networks (WBSN) have made it possible for the development of innovative medical applications to improve health care and the quality of life. Electroencephalography (EEG)-based applications lie at the heart of this promising technologies. However, excessive power consumption may render some of these applications inap- plicable. Hence, intelligent energy efficient methods are needed to improve such applications. In this work, such improved efficiency can be obtained by utilizing smart compression techniques, which reduce airtime over energy-hungry wireless channels; In partic- ular, discrete wavelet transform (DWT) and compressive sensing (CS) are used for EEG signals acquisition and compression. To achieve low-complexity energy-efficient system, the proposed technique makes use of the receiver feedback signals in order to switch between both algorithms based on the application needs. Experimental study has shown that the proposed algorithm effec- tively reconfigures the utilized compression algorithm parameters based on a channel feed back signal. Index Terms—EEG wireless transmission; wavelet compres- sion; compressive sensing; sparse reconstruction algorithms; classification. I. I NTRODUCTION W ireless body sensor networks (WBSN) technologies have a great potential to offer convenient solutions for some medical problems in cost-effective ways. WBSN addresses in particular the mobility problem while the patient is under medical supervision. It outfits patients with wear- able, miniaturized wireless sensors that are able to measure various health related parameters and report them to tele- health providers [1]. Brain event information is often captured by the physiological Electroencephalography (EEG) signals which are extensively used for study of the health status and different activities of the human brain. Recently, wireless EEG sensors are poised to enable the ambulatory monitoring of chronic patients. This imposes some technical challenges regarding the used compression tech- niques in order to provide low transmission rate with accept- able distortion. Many encoding algorithms have been reported in the literature, most of which do not take channel variations into consideration. An algorithm for EEG signal compression was developed in [2], which has shown a good balance between compression ratio and residual distortion, however, This paper was made possible by a NPRP grant 09 - 310 - 1 - 058 from the Qatar National Research Fund (a member of The Qatar Foundation). The statements made herein are solely the responsibility of the authors. it poses a significant increase in computational complexity. In [3], the authors have modified Gabor frame to a chirped Gabor dictionary to further reduce the sparsity level and the corresponding number of measurements, however it does not take channel distortion into consideration. A system modeling framework that quantifies the required power overhead for the compression system was proposed in [4]; however, it did not show which basis functions and compression ratios should be used in order to minimize the reconstruction error. The work in [5] has introduced the use of compressed sensing algorithms for data compression in wireless sensors in order to address the energy and telemetry bandwidth constraints common to wireless sensor nodes. The contribution of this paper can be summarized as follows. An adaptive compression algorithm that switches between the DWT and CS compression algorithms based on the application demands and the hardware specifications. At run-time, the proposed algorithm reconfigures the compression ratio of the utilized compression paradigm according to the channel state such that minimizing the total distortion. The rest of the paper is organized as follows. Section II describes the system model. Both DWT and CS compression techniques are reviewed in section III. Section IV proposes quantitative performance analysis for both compression tech- niques; it also introduces the proposed optimization schemes and explains how they achieve the optimal performance. Section V presents the simulation setup and results. Finally, section VI concludes the paper. II. SYSTEM DESCRIPTION The general structure of the smart wireless monitoring sys- tem is illustrated in figure 1. Here, the distortion is measured by the Percentage Root-mean-square Difference (PRD) as follows, P RD = ‖x - x r ‖ ‖x‖ × 100 (1) where, x is the original signal, and x r is the reconstructed signal. The compression ratio (CR) is evaluated as, CR =  1 - M N  × 100 (2) The 2013 Biomedical Engineering International Conference (BMEiCON-2013) 978-1-4799-1467-8/13/$31.00 ©2013 IEEE