An Asynchronous Multi-Sensor Micro Control Unit for Wireless Body Sensor Networks (WBSNs)

1. Introduction

Recently, the necessity of health monitoring systems using wireless transmission has become prevalent [1–4]. Wireless transmission of health monitoring system data provides comfort and ease in identifying the health conditions of each individual. One of the most practical methods of wireless transmission in health monitoring systems lies on the area of biomedical application in which a wireless body sensor network (WBSN) system is applied. The WBSN system has proven to be an effective method of monitoring health information. In WBSN, the information can be grouped from one wireless body sensor node to another [5,6].

Nowadays, portable wireless body sensor nodes are being developed using a single sensor that can detect only one kind of signal. However, biomedical signals often take different forms such as body temperature [7], heart rate, blood pressure [8], ECG [9], pH value [10], etc., and monitoring such various signals simultaneously would require more than a single body sensor node. The concept of multi-sensor design, which combines different wireless body sensors for flexibility, multi-application, efficiency, and low cost, has been implemented in the literature [11–13]. However, the implementation of multi-sensor architecture imposes a number of design considerations. First, since each sensor handles different kinds of body signals, in order to prevent the occurrence of handshaking problems, the speed of the sensing sensor must synchronize to the clock of the controlling unit. Second, due to the increasing quantity of data from the sensing sensor, which causes bandwidth limitations in the RF module during transmission, the processed data needs some data compression technique to accommodate the whole information. Third, due to the sensitivity of biomedical signals in which the integrity of the information is key, an error correction technique must be implemented. Lastly, since portable wireless body sensor nodes require power to deliver accurate measured result, in order to accommodate a number of body sensor by consuming less power during normal operation, an appropriate power-management supply circuit is needed for power efficiency [14].

Existing general purpose MCUs [15] and DSPs [16] such as the 8051 based controller and C55x are available for possible implementation of the functions discussed above, respectively. In MCU integration design, Amirtharajah [17] presented a micro-power programmable DSP which uses variable precision arithmetic and low-voltage circuits to implement a biomedical detection. Wang [18] designed an 8-bit MCU for a small vocabulary, speaker independent, discrete word speech recognition system. Lee [19] described a smart thermal sensing chip with a MCU design which provides auto-calibration, data compression, and power control for biomedical application. Chen [20] presented a MCU circuit with adaptive low power design for wireless body sensor network. Arm [21] presented a low power design by customizable architecture and reconfigurable instruction set for programmable DSP/MCU core named icyflex1. The mentioned technical works provide great contributions in MCU design, and offer efficient solution for single-sensor of wireless healthcare monitoring systems.

Accordingly, a novel micro control unit is created for wireless multi-sensor body sensor nodes. A novel asynchronous interface is developed to improve the compatibilities of the system which will be composed of devices with different clock domains, which will expand the use range of the wireless body sensor nodes. To increase the detection abilities of each sensor node, we added a multi-sensor control function design to use a single sensor node to detect various biomedical signals by hardware sharing technique. The most important concept in WBSNs is power management. For this reason, a power management technique is developed for reducing power consumption, and furthermore the lossless data encoder is designed to reduce transmission power. Since the reliability is significant in biomedical applications, we developed an error correction coder to prevent loss error from wireless transmissions.

In order to accommodate the above-mentioned necessary functionalities in implementing multi-sensor architecture in biomedical applications, a VLSI implementation of asynchronous multi-sensor micro control unit for WBSNs, which performs more function and control abilities, is proposed in this paper. The proposed architecture is dedicated solely for the multi-sensor application and thus it is low-cost. The control unit of the proposed architecture serves as the core circuit for handling all the processing along with different interfaces. To resolve the issue of handshaking, due to the different clock domains of each sensor, the asynchronous architecture is proposed to ensure proper synchronization. Moreover, the application of the data encoder is included to handle the reduction of transmitted data, which leads to efficient transmission. Lastly, power management and error correction coder (ECC) [22,23] techniques are integrated into the system to minimize power loss and ensure data integrity, respectively.

2. System Architecture

Driven by the growing aging population, the prevalence of chronic diseases, and continuously rising healthcare costs, the wireless body sensor network system is undergoing a fundamental transformation, from the conventional hospital-server system to an individual-medical system. Health monitoring systems can be used to collect health data at home, in the hospital, in outdoor scenarios and even abroad. Figure 1 shows the description of a typical wireless body sensor network system that can be used in applications involving biomedical information acquisition, signal processing, data compressing, and wireless transmission. In this figure an individual (i.e., patient) carries various types of sensors such as heart rate (HR), blood pressure (BP), electrocardiogram (ECG), electroencephalogram (EEG), and body temperature. To be able to accurately collect the multiple health information of the individual in a rapid fashion with utmost data integrity, a special kind of micro control unit [24] is needed.

A block diagram of a wireless multi-sensor node for WBSN system is shown in Figure 1. It consists of sensor modules, a multiplexer for data sensor selection, an analog to digital converter (ADC) module, a micro control unit (MCU), a radio frequency (RF) transceiver, and an antenna. The data being sensed from the sensors are transported into ADC, through a multiplexer and fed into the MCU for processing. In this architecture, the data selection is detected simultaneously and the way it is performed is manipulated by the multiplexer, which is controlled mostly by the MCU. Finally, the processed data is transmitted by the antenna through an RF module and stored in a personal computer server for 24-hour display diagnostic.

Furthermore, the MCU is designed to handle two different signal paths; control path and data path. As the MCU is a commander of a WBSN node for control path dominating, with the sensors, the multiplexer, the ADC, and the RF transmitter with antenna are all controlled in orderly fashion in finite state machines (FSM) for power saving in power management (PWM). The various sensors are governed by the control path of the MCU. The sensed data are condensed to process by the data path of MCU.

3. MCU Control Path

A wireless body sensor node is formed with the sub-circuits of sensors, ADC, RF and MCU. To group each sub-circuit into a single sensor node, a novel controller is created to improve the system performance and compatibility by asynchronous control, reduce power consumption by power management control and save hardware costs by multi-sensor control. In addition, a finite state machine is designed to control each state to ensure a high performance outcome. The MCU includes a central controller for controlling each module of the wireless body sensor node. Moreover, all functions including multi-sensor, ADC, DE, ECC, and RF of the wireless sensor node are controlled by the central controller. Each module of the system is driven by a power source and function control signals from the MCU. A finite state machine (FSM) is designed to detect control states of the ongoing activity of each sensor node. This design enables the system to automatically compute when the user tries to input the sensor signals through the interface.

As shown in Figure 2, the main FSM of the MCU unit consists of seven states. When the adaptive status lies in the zero value, the working state is activated, which means all functions or state conditions of the whole system are powered on. On the contrary, when the status goes high, the normal flow of state signals is activated, by which each state such as sensor, ADC, DE, ECC, and RF functions stay in off condition until the previous state has finished its function. Following Figure 2, a detail illustration with each state represented by its actual device is discussed in Figure 3.

4. MCU Data Path

To decrease the transmission data, the data encoder for lossless compression of various biomedical signals is designed. In addition, an error correct coder circuit is also created for the proposed multi-sensor wireless body sensor network application. By the creative combination of the data encoder and error correct coder, the MCU design gains both benefits of lower data rate and higher communication reliability. Figure 5 shows the data flow of the wireless body sensor node. Since biomedical signals are quite slow (i.e., 1 to 100 Hz [25]), the variations of the adjacent values of body signals are very small. For a 12-bit ADC used for temperature signal from the human body, the resulting values are way too large to be stored or to be transmitted. After the signal passes through the predictor module, the difference values extracted from the original temperature data are measured close to zero. Base from this characteristic, the concept of Huffman coding [26] was chosen to handle the data compression algorithm. The probability of Huffman encoding will be decided by the characteristics of different biomedical signals. Furthermore, the ECC is included for particular characteristics to make sure that 100% of data is correctly transmitted in RF transmission. After transmission through the RF transmitter and reception by the RF receiver, the encoded data stream will be decoded by ECC decoding, entropy decoding, and prediction refining.

5. Experimental Section

In order to realize the WBSN node in hardware applications, we emulated the microcontrol unit design on a FPGA board. The MCU design was implemented by verilog code, simulated by ncverilog tool, and synthesized by design compiler tool. Before the layout of the MCU has been realized by SoC-encounter, the design was initially emulated in Altera FPGA EP2C70F896C6 core. This design was implemented by TSMC 0.13-μm CMOS generic logic process technology. The core area is measured in 0.014 mm² with a power consumption of 496 μW operating at 133 MHz frequency. Figure 10(a) displays the layout photograph of the MCU.

Table 3 lists the specifications of this work and other proposed MCU designs. Including a micro-power DSP design [17] synthesized by 0.6-μm process or MCU design for biomedical application [20] by 0.18-μm process are both different from this work. First of all, a multiple sensors function is performed in this work. For comparison purposes, we convert its core area to 0.13-μm, which allows us to gain a normalized area corresponding to a 0.13-μm process. As a result of normalization, the normalized areas of [17,20] or [21] are 70.9, 4.96, or 31.85 times of this work, resepectively. This work achieves 133-MHz operating frequency, higher than 1.2-KHz, 8.2-MHz, 75-MHz, and 100-MHz of other works. The gate count of this work is 2.68-K, much less by 80% than other precise MCU cores. In order to show power consumption with different operation frequency, the power consumptions of this work operating in 1-MHz, 75-MHz, 100-MHz, and 133-MHz are 3.7 μW, 280 μW, 375 μW, and 496 μW.

Figure 10(b) shows the FPGA emulation of MCU. This system consists of an RF (RS232 protocol) and a FPGA development board. The MCU is compiled by Quartus II and downloaded into ATERA DE2-70 FGPA. The FPGA receives the imported MIT-BIH ADC 11-bit data and process it by the compress and ECC functions of the MCU which then downloaded it into the FPGA. The compressed data from the FPGA output is transmitted from the RF port by the RS232 protocol. After the computer receives the transmitted data, the decoded program, which is written in C language, is decrypted by the decoding algorithms. The result of the decryption is guaranteed to be a 100% correct and refined MIT-BIH ECG database [28].

Table 4 shows the comparison result of different functions of a wireless body sensor node using the MIT-BIH ECG database. For nomal node design including ADC (ADS7842) and RF tranceiver (Zigbee CC2430) [27], the power consumed is 4.5 mW and 81 mW, respectively. Further, the produced wireless transmission error rate is given as 1% bps. In order to obtain the contribution of each creative part, the proposed MCU design was implemeted with creative functions. The WBSN system with power management, DE, and ECC creative function modes and without any mode design are listed in Table 4. The benefit of including the data encoder (DE) not only reduces its data rate but also describes the trasmission data, which means that part of transmitting power is saved. In this case, the creative funtion mode includes power management and data encoder (PWM + DE) that reduced power comsuption by 97.8% and decreased the data rate by 55% more than without the MCU design. Moreover, the ECC design is created to prevent RF transmission errors. In this case, although the ECC creative function mode increases the data rate to 4,620 bit/s, the data rate is still less than that without MCU design with 4,640 bit/s due to the benefit of the data encoder. Futhermore, the creative ECC design can efficeintly reduce 1% bps error rate in this case, which improves the reliability for wireless biomedical applications significantly. To further improve the performance, in this work mode combined with creative DE, and ECC functions, it guarantees that the gains of each creative function are fully utilized.

5. Conclusions

In this paper, we present an asynchronous micro ontrol unit for wireless body multi-sensor networks. This design provides multiple functions and advanced system quality. The operating frequency of the MCU is 133 MHz, with a power consumption of only 496-μW. As compared with wireless sensor nodes without the presence of MCU, this work has the benefits of lower power consumption, less error rate, and faster data rate. These characteristics are the basic requirements for developing a wireless body sensor network (WBSN) system.