Energy-Aware Cross-Layer Optimization for EEG-based Wireless Monitoring Applications Alaa Awad 1 , Ramy Hussein 1 , Amr Mohamed 1 , and Amr A. El-Sherif 1,2 1 Department of Computer Science, Faculty of Engineering, Qatar University, Doha, Qatar 2 Department of Electrical Engineering, Alexandria University, Alexandria 21544, Egypt E-mail: aawad; rhussein; amrm@qu.edu.qa, and amr.elsherif@ieee.org Abstract—Body Area Sensor Networks (BASNs) for healthcare applications have gained significant research interests recently due to the growing number of patients with chronic diseases re- quiring constant monitoring. Because of the limited power source and small form factors, BASNs have distinguished design and operational challenges, particularly focusing on energy optimiza- tion. In this paper, an Energy-Delay-Distortion cross-layer design that aims at minimizing the total energy consumption subject to data delay deadline and distortion threshold constraints is proposed. The optimal encoding and transmission energy are computed to minimize the total energy consumption in a delay constrained wireless body area sensor network. This cross-layer framework is proposed, across Application-MAC-Physical layers, under a constraint that all successfully received packets must have their delay smaller than their corresponding delay deadline and with maximum distortion less than the application distortion threshold. Due to the complexity of the optimal-proposed solu- tion, sub-optimal solutions are also proposed. These solutions have close-to-optimal performance with lower complexity. In this context, there is complexity/energy-consumption trade-off, as shown in the simulation results. Index Terms—Wireless healthcare applications, EEG signals, BASNs, Convex optimization, Cross-layer optimization. I. I NTRODUCTION The rapid increase in the number of people living for years with chronic conditions, that require ongoing clinical man- agement, has increased the importance of electrocardiogram (ECG) and electroencephalogram (EEG) diagnosis systems. Advances in wireless sensing and wearable sensors have made body area sensor networks technology a promising solution, to meet this growing demand, and surpassing opportunity for ubiquitous real-time healthcare monitoring without con- straining the activities of the patient [1]. Wireless body area sensor networks consist of tiny nodes in, on, or around a human body to monitor vital signs such as body temperature, activity or heart-rate. These sensor nodes periodically send sensed information to a coordinator node. To reduce energy consumption, it is assumed that all these sensor nodes are in standby or sleep mode until the centrally assigned time slot. There is no possibility of collision within the network, as all communication is initiated by the central node and is addressed This work was made possible by NPRP grant # 09-310-1-058 from the Qatar National Research Fund (a member of Qatar Foundation). The statements made herein are solely the responsibility of the authors. uniquely to each node. The conventional wireless sensor network technologies are typically bulky, power hungry and based on MAC protocols such as Bluetooth and Zigbee/IEEE 802.15.4, which are inefficient for such BASNs applications [2]. They also ignore the cross-layer design which optimizes the performance by jointly considering multiple protocol lay- ers. In the past few years, much of the research in the area of BASNs has focused on issues related to wireless sensor designs, sensor miniaturization, signal compression techniques and low-power hardware design [3][4][5][6]. A good review of state-of-the-art hardware, technologies, and standards for BASN was presented in [7]. To the best of our knowledge, the cross-layer design of energy minimization to address distortion constraints for delay sensitive transmission of EEG traffic in BASNs has not been studied before. For example, the authors in [8] investigate the properties of compressed ECG data for energy saving using a selective encryption mechanism and a two-rate unequal error protection scheme. Other researches focus on reducing power consumption at MAC layer by avoiding idle listening and collision [9], or by presenting latency-energy optimization [10]. The authors in [11] developed a MAC model for BASNs to fulfill the desired reliability and latency of data trans- missions, while simultaneously maximizing battery lifetime of individual body sensors. For that purpose, a cross-layer fuzzy-rule scheduling algorithm was introduced. However, they ignored the encoding energy and source coding distortion in their model. Security of BASNs also becomes one of the attractive research points [12], due to medical data regulations. In our model, the sensor nodes transmit the sensed in- formation with dynamic power adaptation technique to their coordinator. This is because the wireless link quality can change rapidly in body area networks, and a fixed transmit power results in either wasted energy (when the channel state is good) or low reliability (when the channel state is bad) [13]. This model is distinct from [14], which assumes that the sensor nodes transmit their information at a constant power to their coordinator. Furthermore, the authors in [14] did not take the source coding distortion nor the encoding energy into consideration. In this paper, to anatomize, control, and optimize the behavior of the wireless EEG monitoring system under the 38th Annual IEEE Conference on Local Computer Networks 978-1-4799-0537-9/13/$31.00 ©2013 IEEE 356