Low-power design considerations to enable effective health and fitness bio-patch solutions
Bio-patch solutions are sensors worn on the body to enable continuous (or semi-continuous) monitoring of physiological and cognitive parameters without tethering the patient or athlete to a wired hub. They are poised to revolutionize the health and fitness market and create new ways of providing healthcare in clinical and remote settings.
Given the unobtrusive and small form requirements of the bio-patch, optimizing power efficiency becomes highly critical in order to extend the lifetime of the system.
Introduction to bio-patch solutions
Bio-patch solutions have the ability to monitor both the physiological and cognitive functions for an extended period of time through a wireless gateway, outside of a clinical setting allows for innovative health management solutions.
The wearable nature of the bio-patch enables more intimate skin contact compared to other reusable wearable solutions providing for more accurate data collection while the disposable factor helps meet patient safety requirements in hospital settings.
Bio-patches can be used to monitor a number of physiological parameters ranging from simple on-skin temperature measurements to more sophisticated electrocardiogram (ECG) type measurements. In Table 1, we list a number of sensor solutions with corresponding parameters that can be quantitatively measured. The sensor solutions can be classified into two main categories consisting of physical sensors and chemical sensors.
System low-power factors
Expected battery lifetimes in bio patches range from 12-24 hours in clinical settings where the raw data is continuously transmitted to 7-10 days in a home-health or sports and fitness setting where the data is periodically transmitted. Those battery lifetimes can only be achieved by optimizing the energy efficiency of the entire system. Extended run-time can also be obtained by using an energy harvesting scheme. A systems view of the bio-patch includes: RF interface and embedded processing requirements, sensor data collection subsequent signal conditioning, and power management.
An example block diagram of the bio-patch solution is shown in Figure 1. In most wireless systems, the RF component tends to drive the overall power efficiency of the solution if not optimized for the use-case condition.
RF interface and embedded processing requirements
There are two use cases for bio patches:
- Solution transmits raw data to the hub where the data is processed and displayed, and the signal conditioning algorithm resides on the hub with a large RF duty cycle (Figure 2).
- Sensor signal conditioning algorithm resides on the sensor itself and only transmits the required information during predefined periods of time, transmitting only packets of information that correlate to a change in the physiological parameter being monitored.
The composite energy consumption for the system is given by the sum of the energy consumption for each of the major components in the system for the given RF protocol used for communication:
Ecomposite = EMCU total + Esensor total + Emem prog + Emem erase + ERF (1)
ERF = Elisten + Et + Er + Esleep (2)
EMCU total is the total energy consumption for the microcontroller (MCU) which consists of the sum of the active, idle and switching components, Esensor total is the sum of the power consumption for each of the sensors, Emem prog is the amount of energy required to carry out data logging and Emem erase is used to account for Flash block erase requirements associated with writing to this memory technology. ERF is determined by the RF protocol used in the communication channel. For the Phy and MAC layers of an IEEE802.5.4 protocol, Elisten is the active listening energy, Et is the energy for packet transmission, Er is the receive energy and Esleep is the radio sleep energy. The lifetime of the end node is dependent on the total energy consumed by the system and the battery capacity. The end node lifetime is determined as follows:
Lnode lifetime = [ Cbatt x V ] / Ecomposite (3)
where Cbatt is the battery capacity used in the system.
One way to maximize the overall battery lifetime of the bio-patch solution is to minimize the RF duty cycle. As a comparison, we run an analysis using the device parameters shown in Table 2.
A number of RF standard protocols have emerged which have been designed to minimize the RF duty cycle by optimizing for occasional connections that allow for longer sleep times between connections and small data transfers. Bluetooth low energy (BLE) and Near-field communication (NFC) for example. Eliminating the RF component, translates to orders of magnitude increases in the overall battery lifetime of the bio-patch. At the lower RF duty cycles, the other components of the system have a more prominent role in the overall power efficiency. This is highlighted by the lower expected battery lifetime of the system with the MCU running at a higher active current, as shown in Figure 3.
Data logging and battery life
Now let’s look at the impact of data-logging on the overall battery lifetime of the system using two different non-volatile memory technologies – Flash versus ferroelectric random access memory (FRAM). Each individual bit can be accessed, and unlike EEPROM or Flash, FRAM does not require a special sequence to write data, nor does it require a charge pump to achieve the higher programming voltages. FRAM programs at 1.5V versus the 10-14V of Flash or EEPROM.
In addition, FRAM is about 1,000 times faster than the previously mentioned nonvolatile counterparts. Because the speed of FRAM is equivalent to embedded static RAM in many MCUs, in addition to its dynamic accessibility and non-volatility, it is what is commonly referred to as a “universal memory.” This means it can function as the data memory or the program memory at any given time in its life, giving designers the freedom to create embedded software that either relies heavily on data processing or does not rely at all on data processing, depending on their specific needs without worrying about the limitations of the MCU.
Flash is limited to approximately 10,000 write cycles while FRAM write cycles are in the billions. FRAM can be used in true data-logging applications where data needs to be retrieved when system power is lost. To define the difference in the energy efficiency of the two memory technologies, we use the device parameters listed in Table 3 and calculate the battery lifetime of a bio-patch where the sensor data is logged, accounting for the erase cycle of the Flash memory in the calculation and maintaining the RF duty cycle.
As can be seen in Figure 4, data-logging using FRAM does not impact the overall battery lifetime even for the case where the sensor is collecting 32 bytes of memory for the given sensor cycle. But Flash results in a significant drop in the battery lifetime of the bio-patch – up to 30 percent for 32 bytes of sensor-collected data.
Signal chain and conditioning of the bio-patch
To finalize the signal chain of the bio-patch solution we look at the impact of the signal conditioning on the overall power efficiency calculation. Driving the adaptive signal conditioning to the analog front end prior to the analog to digital conversion (ADC) reduces the computational requirements of the MCU and also minimizes the on time of the processor. As an example, we take a look at the sensor response of a typical electrocardiogram (ECG) signal shown in Figure 5. The superimposed transients drive a requirement for an ADC with a higher resolution had the transients been removed prior to sensing with the ADC. The power performance of a 14-bit successive approximation (SAR) ADC is significantly better compared to a 22-bit SAR ADC.
An optimization of the power efficiency is carried out by an understanding of the different system components that make up the total signal chain of the bio-patch solution in relation to the specific use case. For continuous monitoring solutions, we see that the RF component drives the overall system lifetime. For semi-continuous or ‘on-demand’ solutions with lower RF duty cycles, the other components of the signal chain contribute a significant share of the complete power efficiency breakdown.