The healthcare industry has been adopting greater use of digital technology over the last couple of decades. The COVID-19 pandemic helped accelerate this evolution. The remote access to healthcare necessitated by the pandemic highlighted several other benefits, such as more efficient healthcare delivery and continuous patient monitoring. Technological advances have spawned the Internet of Medical Things (IoMT), where networks of patients with portable and/or wearable medical devices and sensors, and the corresponding healthcare systems and providers, are connected through the Internet. Continuous blood glucose levels and heart monitors are examples of devices that have gained widespread acceptance. IoMT devices help automate data transfer, thereby reducing human error. Advances in predictive data analytics and artificial intelligence (AI) make IoMT devices even more powerful by enabling data-driven diagnostics with early detection of abnormalities, greater patient self-engagement, and reduced healthcare costs.
Key requirements for IoMT devices
- Security: The sensitive nature of the medical information being transferred requires a high level of security. The Advanced Encryption Standard (AES) and Elliptical Curve Cryptography (ECC) can encrypt and decrypt data transfer using secure keys and hence authenticate the data. Keys based on a true random number generator (TRNG) in the device help in the secure generation of these keys. Spoofing attacks can be minimised with the use of device identification using unique physically unclonable functions (PUF) within the semiconductor device. Secure boot-up hardware protocols, as well as tamper-proof mechanisms that prevent access to protected regions of the device memory help enhance device security.
- Power consumption: Wearable and portable devices typically run on battery power. Low-power communications protocols such as Bluetooth LE 5.x, power-saving modes when the device is not active, and an efficient architecture that optimises operational performance versus power consumption are some essential features that can maximise battery life.
- Rich feature set in a small size: Small and light devices enable their use in wearable and portable medical applications. New applications such as smart tooth implants require tiny form factors. The System on Chip (SoC) concept provides a high level of multi-functional integration onto a single chip. This can include a peripheral feature set that provides high-speed analogue and digital sensing, measurement, data transformation, and communication. Other essential requirements include wireless connectivity, high-speed data processing with large flash and RAM memory, precision low-frequency/low-power clocks and timers, DC/DC voltage regulation, etc.
Silicon Labs EFRBG27 wireless Gecko SoC family for IoMT applications
In March 2023 Silicon Labs announced the release of a new family of secure, energy friendly devices that expand their Wireless Gecko portfolio. This includes the BG27 series of Bluetooth LE SoC devices ideally suited for IoMT applications.
A block diagram showing the rich feature set included in the BG27 SoCs is shown in Figure 1. Some details on the key features are listed below:
Processor and memory:The 76.8 MHz, 32-bit ARM Cortex M33 RISC core with DSP instruction and floating-point unit allows for high-performance signal processing capability at 1.50 Dhrystone MIPS/MHz. It includes the ARM TrustZone security technology. The flash memory is 768 kB, while the data memory is 64 kB of RAM. The Linked Direct Memory Access Controller (LDMA) allows the system to perform memory operations independent of the software, hence reducing energy consumption and software workload.
Low power modes: The EFR32BG27 includes an Energy Management Unit (EMU) that manages transitions of the energy modes (EM0 to EM4) of the SoC. With the EMU, applications can dynamically minimise energy consumption during program execution. EM0 mode provides the highest number of features, such as enabling the CPU, radio, and peripherals at the highest clock frequency. Peripherals can be disabled in the low-power active modes EM2, EM3. Voltage scaling is used by the EMU when transitioning between energy modes to optimise energy efficiency by operating at lower voltages when possible. EM4 is an inactive, lowest power state that allows the system to wake up into the EM0 mode.
DC/DC conversion: The EFR32BG27 family includes both buck and boost mode on-chip converters that can supply the required internal 1.8 V. The boost mode devices, such as the EFR32BG27C230F768IM32-B, have the ability to operate down to 0.8V, allowing single cell alkaline, silver oxide, and other low-voltage battery operation. The boost converter can be shut down using a dedicated BOOST_EN pin, hence saving system battery power during storage and shipping. In this mode, the maximum current draw is only 20/50nA, depending on the powering of certain pins. In the buck mode devices, such as the EFR32BG27C140F768IM40-B, a maximum 3.8 V can be supplied externally. An on-chip supply monitor signals when the supply is low enough to allow the regulator to be bypassed and extend the range to 1.8 V. The bypass mode also allows the system to go into the EM4 energy-saving mode. A Coulomb Counter block is integrated into the DC/DC converter. This includes two 32-bit counters that are used to measure the number of charge pulses delivered by the DC/DC converter, enabling accurate battery level tracking to enhance user safety.
Bluetooth 5.x networking: Bluetooth Low Energy (LE) wireless protocol is supported by this SoC family. The radio receiver uses a low-IF architecture consisting of a low-noise amplifier and an I/Q down conversion. The automatic gain control (AGC) module adjusts the receiver gain to avoid saturation for improved selectivity and blocking performance. The 2.4 GHz radio is calibrated at production to improve image rejection performance. The family includes a range of transmit powers from 4 dBm to 8 dBm. RF noise mitigation includes operation of the DC/DC converter in soft switching mode at boot, and DC/DC regulating-to-bypass transitions to limit maximum supply slew rate and mitigate inrush current. The RFSENSE block allows the device to remain in EM2, EM3 or EM4 energy-saving modes and wake up when RF energy above a specified threshold is detected.
Security: The EFR32BG27 family of SoCs includes a range of security features, as shown in Figure 2.
Figure 2: Security features of the EFR32BG27 wireless Gecko SoC family. (Image source: Silicon Labs)
The Secure Boot with the Root Of Trust and Secure Loader (RTSL) authenticates trusted firmware that begins from immutable read-only memory (ROM). The cryptographic accelerator supports AES and ECC encryption and decryption. It also includes Differential Power Analysis (DPA) countermeasures to protect keys. The TRNG harvests entropy from a thermal source and includes start-up health tests for this source, as required by the NIST SP800-90B and AIS-31 standards, as well as online health tests as required for NIST SP800-90C. The debug interface, locked when the part is released in the field, has a secure unlock function that allows authenticated access based on public key cryptography. On the hardware side, an External Tamper Detect (ETAMPDET) module enables detection of external tampering such as unauthorised enclosure opening. It can generate an interrupt to warn the software and allow system-level actions to be taken.
Rich peripheral set: The SoCs include hybrid analogue-to-digital converters that combine both SAR and Delta-Sigma techniques. The 12-bit mode can operate at speeds of up to 1 Msps, while the 16-bit converter can operate at up to 76.9 ksps. The analogue comparator module can use internal or external references and can also be used to sense the supply voltage. SPI, USART and I2C serial communication modes are all supported. The Real Time Clock and Capture (RTCC) module provides 32-bit timekeeping down to EM3 power modes and can be clocked with the internal low-frequency oscillator. The Low Energy Timer (LETIMER) provides 24-bit resolution and can be used for timing and output generation when most of the device is powered down, allowing simple tasks to be performed with minimal power consumption. The Peripheral Reflex System (PRS) is a signal routing network that allows direct communication between the peripheral modules without involving the CPU. This reduces software overhead and current consumption.
Low footprint packages: One of the devices in the EFR32BG27 family is the EFR32BG27C320F768GJ39-B. This device comes in a wafer-level chip scale package (WLCSP) with dimensions of just 2.6 mm x 2.3 mm and can be run in either buck or boost regulator modes. The rest of the family comes in either QFN32 4 mm x 4 mm, or QFN40 5 mm x 5 mm packages in specific regulator modes of either buck or boost.
The EFR32BG27 provides industry-leading energy-efficient processing capability and low-energy Bluetooth connectivity. These small form factor SoCs, that include a variety of security features, are ideally suited for IoMT applications.
Rolf Horn, Applications Engineer at DigiKey, has been in the European Technical Support group since 2014 with primary responsibility for answering any Development and Engineering related questions from final customers in EMEA, as well as writing and proof-reading German articles and blogs on DK’s TechForum and maker.io platforms.