ULP Wireless Update

Benchmark reveals efficiency of Bluetooth low energy ‘end nodes’

Benchmark reveals efficiency of Bluetooth low energy ‘end nodes’

EEMBC’s IoT Connect framework comprises an energy monitor, a radio manager to coordinate communication, and an IO manager to synchronize activities and to simulate a sensor input

Industry-backed benchmark allows developers to directly compare Bluetooth low energy SoC power consumption in wireless ‘thing’ applications connected to the IoT

Low power wireless ‘things’ dutifully monitoring their surroundings at the very edges of the IoT are likely to reside in inaccessible places. That makes it important that their batteries last a long time because changing them might be inconvenient and time consuming.


Bluetooth low energy, as the name suggests, was designed to minimize power consumption and modern SoCs from a range of manufacturers perform well, typically drawing milliamp peak currents and microamp average currents. Such modest power consumption enables batteries to last for months in low duty-cycle applications. But some chips are better for certain applications than others, and even a few percentage points difference in performance can extend battery life by weeks.


Silicon vendors include power consumption data for a range of Bluetooth low energy radio modes in their data sheets, but these are often measured in optimum conditions for limited applications. That can make it difficult for a developer to make an unbiased comparison between shortlisted components for his or her specific application based on the data sheets alone.


Benchmarks to the rescue

Now the Embedded Microprocessor Benchmark Consortium (EEMBC) has come to the rescue. The organization, formed in 1997 to develop industry-standard benchmarks for embedded system hardware and software, has recently announced the availability of EEMBC IoTMark-BLE, a benchmark and analysis tool that measures the energy efficiency of Bluetooth low energy SoCs used in IoT ‘end nodes’. (The organization defines an end node as comprising four parts: sensor; processing (including the RF protocol firmware); interfaces connecting sensor and processing; and a communication mechanism used to transmit/receive information.)


“Datasheets ... tell only part of the story,” said Brent Wilson, co-chair of the EEMBC IoT-Connect working group, in a statement. “[The new benchmark] measures the energy used by the full [radio] system ... while it performs real-world tasks. Each part of the system impacts energy efficiency, so evaluating the whole radio system provides the most realistic evaluation of its battery life.”


The organization says the benchmark allows developers to make comparisons between embedded microcontrollers, radio physical layers, and modules, as well as between RF protocol firmware. The framework supporting the benchmark includes an energy monitor, a radio manager to coordinate the communication with the device under test (DUT), and an IO manager to synchronize activities and to simulate a sensor input on the DUT’s I2C or serial peripheral interface (SPI).


Applications determine a range of parameters including packet payload, frequency of packet transmission, and transmit power. EEMBC members agreed on specific tests and measurements to cover a wide range of applications.


EEMBC claims the framework “makes it easy for developers to select parameters and functions to ensure equitable comparisons while providing valuable insight beyond datasheet specifications”.


The Bluetooth low energy benchmark is the first of several IoT wireless technology comparisons planned by the EEMBC’s working group. The clever part of the organization's strategy is that the benchmark framework can be applied to other RF communication protocols allowing new benchmarks to be rolled out quickly. Next up are benchmarks for Wi-Fi, 6LoWPAN, and Low Power Wireless Area Networks (LPWANs).


“The framework ... has the flexibility to accommodate various communication protocols, and is also sufficiently portable to work with any vendor’s microcontroller and radio-module products,” said Peter Torelli, EEMBC director of technology, in a statement. ”Industry-standard benchmarks [lead developers] to informed decisions, rather than having to demystify unsubstantiated claims from competitors.”


The working group includes Nordic Semiconductor, embedded microprocessor IP vendor, ARM, and other major wireless chip manufacturers. EEMBC is encouraging other interested parties to join the group to ensure it covers all envisaged applications.


“Nordic, along with other major RF chip makers, supports the EEMBC’s IoTMark benchmark initiative because it means network developers can choose the best wireless device for their specific application based on impartial comparison against alternatives,” says John Leonard, a Product Marketing Manager with Nordic. “That can only result in the roll-out of higher quality IoT infrastructure.”