Performance Requirements for Wireless Sensor-Based Personal Area Networks in Medical Monitoring



This article looks at the different types of sensors used in wireless personal area networks for medical monitoring and how their data requirements map to the different wireless network protocols available, ranging from BPAN to ZigBee and Bluetooth Low Energy. It covers products such as the Freescale EKG development kit, as well as Bluetooth and ZigBee devices from Texas Instruments, Freescale, and Atmel.

Personal area networks, particularly medical systems, are all about sensors. The aim is to capture data from a range of sensors around the body and carry that information to a place where it can be stored, analyzed and if necessary, acted upon quickly. Using wires to link sensors to a base station has restricted the use of the technology to hospital rooms, limiting the wider use of potentially life-saving systems. Using wireless technology to link the sensors to a base station is opening up new ways of monitoring patients and new opportunities for system development.

These wireless sensor networks (WSN) are changing the way consumers and healthcare organizations think about health and wellness, according to global technology research firm ON World.

In 2017, 18.2 million health and wellness WSN systems — excluding sports and fitness devices — will be shipped worldwide and annual revenues enabled by these systems will reach $16.3 billion. Cloud connected services will make up 53% of the revenues in five years.

By that time, solutions for chronic conditions such as blood glucose management and cardiac/ECG monitoring will make up 60% of the revenues. However, general wellness will increase by 1600% over the next five years when it will make up 41% of the device shipments.

Health cloud platforms combined with sensor networks are also providing bundling opportunities for telehealth, telecare and home service providers, with over half focused on self-management, prevention and general wellness. Smartphone and cellular innovations include diabetes management systems, electrocardiograph (ECG) monitoring system and an NFC blood pressure monitor. A growing number of Bluetooth health products are emerging for use with smartphones for blood glucose monitoring as well as blood pressure monitors.

For example, the Polar T34 Non-Coded Heart Rate Transmitter from Parallax monitors and then wirelessly transmits the heart rate data from the chest strap to a Polar WearLink+ compatible receiver. This allows the wearer to monitor their heart rate and related biometric statistics. This transmitter can be paired with local gym exercise equipment if it is Polar WearLink compatible.

It wirelessly transmits heart rate data without the need for conductive gel and the battery lasts up to 2500 hours of continuous usage.

Figure 1: The Parallax Polar T34 heart rate transmitter.

Because of this growth, there two new wireless protocols emerging that will help with the development of such systems. Bluetooth Low Energy, part of Bluetooth v4.0, is providing quick connections to keep the power consumption down, with the advantage of easily linking to an existing smartphone for storage and analysis and onward transmission. This uses point-to-point connections and so is appropriate for single sensor systems or sensor fusion systems such as the Non-Coded Heart Rate Transmitter shown in Figure 1. For a regularly used Bluetooth network around the body, encryption may have to be considered to protect regular data links from interception, and wireless devices are now including 128-bit AES encryption blocks to easily protect the sensor data.

ZigBee is another wireless protocol that is optimized for wireless sensor networks with low power and short, reliable links that can be used for streaming data from sensors around the body. A key capability of ZigBee is the mesh networking capability. This allows a node to forward data from another node, allowing the links to be significantly shorter. While this does not work well with high data rates, it is ideally suited to streaming sensor data and to medical networks with multiple sensors. For example, an ECG heart monitor may have several field effect sensors at different points on the chest, all feeding to a central hub that collates the data. The field effect sensors need to stream the data back to the hub reliably to capture all the heartbeat information, as small variations in the signal can be significant to the doctor.

This makes it difficult to use a duty cycle to reduce the power consumption of the link, so the shorter link distance means that the power consumption of the RF transceiver that is interfaced to the sensor can be significantly reduced, extending the battery life.

A variation of ZigBee for medical personal networks, 802.15.4j, was approved in February 2013 to use a different set of frequencies. In the US, this is the 2.36 GHz to 2.4 GHz band that sits just below the unlicensed 2.4 GHz band used by Wi-Fi, traditional ZigBee and Bluetooth networks. This modifies the PHY and MAC layers of the wireless link to use these less congested and regulated frequency bands for medical wireless sensor networks. As the links are not subject to interference, they can use lower power and run more reliably, again boosting the battery life for the batteries that power the sensors or allowing energy harvesting sources such as body heat and movement to be used instead of a battery.

Digital stethoscope

One example of a connected medical system is a digital stethoscope that combines a microphone with an analog-to-digital converter to translate the sound of a patients breathing into a digital stream that can be transmitted over a wireless link. The stethoscope analog front-end often provides audio output in selectable modes that determine the frequency response needed by the microphone and the range of the ADC. The Bell mode (20 Hz to 220 Hz) emulates the light touch of a stethoscope to pick up low frequency sounds while the Diaphragm mode (50 Hz to 600 Hz) emulates the firmer contact method for picking up higher frequency sounds. The extended range (20 Hz to 2000 Hz) of course covers the whole range, rather than emulating a traditional acoustic sensor.

The digital stethoscope front end can connect through a standard interface to various processor platforms, such as the CC2540, TI’s cost-effective, low-power, true system-on-chip (SoC) for Bluetooth low energy (BLE) applications. It enables robust BLE master or slave nodes to be built with very low total bill-of-material costs, linking sensors in a medical personal network.

The CC2540 uses an industry-standard enhanced 8051 MCU, in-system programmable flash memory, 8 KB RAM, and very low power sleep modes, ideal for medical networks. Short transition times between operating modes further enable low power consumption so the device can power up quickly, send the data and power back down again, minimizing the energy used.

The 8051 CPU core is a single-cycle 8051-compatible core. It has three different memory access busses (SFR, DATA, and CODE/XDATA), a debug interface, and an 18-input extended interrupt unit.

The memory arbiter is at the heart of the system, as it connects the CPU and DMA controller with the physical memories and all peripherals through the SFR bus, and this allows the sensor data to be streamed into the RF transceiver. The memory arbiter has four memory-access points, access of which can map to one of three physical memories: an SRAM, flash memory, and XREG/SFR registers. It is responsible for performing arbitration and sequencing between simultaneous memory accesses to the same physical memory.

Five channels in the DMA controller provide access to external sensors and to the memory using the XDATA memory space; and from there, to all the physical memories. Each channel (trigger, priority, transfer mode, addressing mode, source and destination pointers, and transfer count) is configured with DMA descriptors that can be located anywhere in memory. Many of the hardware peripherals (AES core, flash controller, USARTs, timers, ADC interface, etc.) can also be used with the DMA controller for efficient operation by performing data transfers between a single SFR or XREG address and flash/SRAM. This simplifies the connection of the sensor to the transceiver via a USART or ADC.

The interrupt controller services a total of eighteen interrupt sources, divided into six interrupt groups, each of which is associated with one of four interrupt priorities, which can come from any one of the sensors connected up to the device. I/O and sleep timer interrupt requests are serviced even if the device is in a sleep mode (power modes 1 and 2) by bringing the CC2540 back to the active mode, keeping the power consumption low.

Data can be delivered to the transceiver via the ADC, which supports 7 to 12 bits of resolution with a corresponding range of bandwidths from 30 kHz to 4 kHz, respectively. DC and audio conversions with up to eight input channels (I/O controller pins) are possible. The inputs can be selected as single-ended or differential. The reference voltage can be internal, AVDD, or a single ended or differential external signal. The ADC also has a temperature-sensor input channel, which can be useful for medical monitoring to simplify the design of the wireless sensor node. The ADC can automate the process of periodic sampling or conversion over a sequence of channels, relieving the designer of the need to code this.

The operational amplifier is intended to provide front-end buffering and gain for the ADC. Both inputs as well as the output are available on pins, so the feedback network is fully customizable. A chopper-stabilized mode is available for applications that need good accuracy with high gain.

Encrypting the data in the medical network is a key decision for the designer. The data links are very low power and very short range and therefore hard to intercept, but ensuring that medical data is secure and safe from hacking is a vital element of any design. ZigBee includes encryption within its protocol, but Bluetooth does not, so there has to be more consideration for implementing data security. The AES encryption/decryption core in the CC2540 allows the user to encrypt and decrypt data using the AES algorithm with 128-bit keys.

The Medical EKG Module (MED-EKG) from Freescale Semiconductor is a low-cost development board that allows users to rapidly prototype electrocardiogram (EKG/ECG) applications. Users can choose to develop EKG/ECG applications using the external analog components or by using the microcontroller’s on-chip analog modules or a mixture of both.

Figure 2: Block diagram for the MED-EKG development board.

Figure 3: The MED-EKG development board from Freescale Semiconductor.

The EKG/ECG software demonstration and step-by-step lab guide is available in the both the Tower System, which will also allow development boards for the wireless part of the design to be added using devices such as Freescale’s third generation ZigBee platform.

The MC1322x incorporates a low power 2.4 GHz radio frequency transceiver with a 32-bit ARM7 core based MCU, hardware acceleration for both the IEEE 802.15.4 MAC and AES security, and a full set of MCU peripherals into a 99-pin LGA Platform-in-Package (PiP).

The 32-bit ARM7TDMI-S core operates up to 26 MHz while the 128 Kbyte flash memory is mirrored into a 96 Kbyte RAM so that upper stack and applications software can be easily implemented. In addition, an 80 Kbyte ROM is available for boot software, standardized IEEE 802.15.4 MAC and communications stack software. A full set of peripherals and Direct Memory Access (DMA) engine support the integration of sensors alongside the transceiver. All components are integrated into the package except the crystal and antenna, allowing the device to be as small and light as possible.

Freescale has two independent codebases to support the two ZigBee standard specifications, but the key for medical sensor network applications is the BeeStack as this supports ZigBee-2007 and ZigBee Pro extensions and will support the 802.15.4j-2013 amendments.

The BeeStack architecture implements the ZigBee-2007 protocol stack including both Stack Profile 1 and Stack Profile 2 (Pro). The PHY, MAC, and network (NWK) layers create the foundation for the application (APL) layers, but BeeStack defines additional services to improve the communication between layers of the protocol stack. The key is that the IEEE 802.15.4-compatible MAC/PHY layer is not part of ZigBee itself. The NWK layer defines routing, network creation and configuration, and device synchronization while the application framework (AF) supports services that define ZigBee functions. The ZigBee Alliance also defines profiles to target specific markets, and there is a profile for healthcare applications.

Atmel similarly has a low-power 2.4 GHz radio transceiver designed for medical sensors using the ZigBee IEEE 802.15.4 protocol. The AT86RF231 can be operated using an external microcontroller to interface to the transceiver, where all the RF-critical components are integrated on-chip with the exception of the antenna, crystal and de-coupling capacitors. This makes the AT86RF231 particularly suitable for small-form-factor low-power MBAN networks, while simplifying the sensor connections to the external microcontroller.

These can use a master SPI interface to link directly to the AT86RF231. The SPI is used for register, Frame Buffer, SRAM and AES access. The additional control signals are connected to the GPIO/IRQ interface of the microcontroller. Pin 17 (CLKM) can be used as a microcontroller master clock source. If the microcontroller derives the SPI master clock (SCLK) directly from CLKM, the SPI operates in synchronous mode, otherwise in asynchronous mode.

Figure 4: Interfacing the Atmel AT86RF321 ZigBee transceiver to a microcontroller via SPI.

In synchronous mode, the maximum SCLK frequency is 8 MHz. In asynchronous mode, the maximum SCLK frequency is limited to 7.5 MHz. The signal at pin CLKM is not required to derive SCLK and may be disabled to reduce power consumption and spurious emissions.

Conclusion

The new ultra-low power protocols will bring more battery life and more flexibility to medical personal networks that use sensors. Integrating the sensors with the single chip transceivers provides a cost effective way to deliver the small size and longer battery life that is needed for such applications, and the choice of Bluetooth or ZigBee for the different application areas and use cases gives the engineer considerable flexibility in optimizing the design.
Supplier