BAN 기반 심전도 모니터링의 에너지 효율과 신뢰성 향상을 위한 온노드 프로세싱 및 우선순위 MAC

Abstract

With the development of wireless communication, sensor design, and energy storage technologies, wireless sensor networks (WSNs) have become a reality. Currently, monitoring human body physiological parameters is an interesting application of WSNs. However, the well-designed WSNs are not ideally suited to these applications. For example, the power consumption of a sensor node should be reduced to the minimum to sustain long term monitoring, and the transmission reliability should be further emphasized with life threatening information delivered through the network. Therefore, it is necessary to develop a body area network (BAN) to meet the needs of these healthcare applications with lower power consumption and reliable transmission. BAN based continuous monitoring systems have great potential for managing people’s chronic health conditions. Many healthcare applications based on BANs have been researched, among which the continuous monitoring of people who are suffering from heart illnesses is the most remarkable. A BAN based ECG monitoring system allows doctors to pervasively monitor patients’ heart conditions in any environment without any restriction of the patients’ activities. Energy efficiency is one of the biggest challenges to the ECG monitoring application of BANs, because the battery powered sensor node has very limited energy capacity. It cannot afford the long term monitoring by continuously sensing and transmitting the ECG signals to the doctor. However, the energy can be used in a more efficient way to prolong the life time of the sensor node. Thus far, there are three main ways for saving energy of the sensor node: low power hardware design, light weight software design, and smart working mechanism. Many efforts have been devoted to designing energy efficient hardware such as low power transceivers, processors, and software such as light weight MAC protocols and routing protocols. However, only a few works, either directly porting existing ECG analysis algorithms or simply measuring duration and amplitude of ECG signals, have been done from the system working mechanism point of view to save power consumption of the sensor node. Our motivation is to use smart working mechanism of the sensor node for saving energy. We put some intelligence on the sensor node by trading computation for communication so that continuous transmission is not necessary all the time. In this work, we thoroughly analyze the power consumption of sensor nodes and setup the framework of an intelligent sensor node based ECG monitoring system. Then, we propose a light weight ECG analysis algorithm for local diagnosis, which achieves a sufficiently high classification accuracy and energy efficiency. The proposed ECG analysis algorithm classifies ECG cycles into normal or abnormal by computing the Euclidean distance between a tested ECG cycle and a normal ECG cycle. We also extend by including a prioritized MAC protocol to improve the reliability of reporting abnormal information. The prioritized MAC protocol is characterized by two features: i) It prioritizes packets by increasing the backoff time of non-urgent packets; ii) The channel is also cleared for the urgent packets by ceasing the transmission of all other non-urgent packets. Moreover, we provide extensive experimental simulation and implementation results for validating the efficiency of these proposed schemes. The local ECG analysis algorithm is evaluated from two aspects－the power consumption and diagnosis accuracy. The prioritized MAC protocol is estimated from three aspects, namely, the average packet delay, the retransmission probability, and throughput. These results demonstrate the energy efficiency and reliability of the intelligent sensor node based ECG monitoring system.