Explaining the SiWx917 Low-Power Wi-Fi Features for IoT Product Developers

About this Whitepaper

The SiWx917 is the world's most energy-efficient Wi-Fi 6 wireless microcontroller unit (MCU) and can deliver years of battery-life for IoT devices. It is equipped with an intelligent power management solution that provides IoT developers more fine-grained energy optimization possibilities than any other product in the market. An independent testing provider estimated the SiWx917 is capable of providing up to 5-years of battery life for smart locks.

However, simply mounting the world's most energy-efficient Wi-Fi 6 chip on your product doesn't bring years of battery-life or low standby current: you need to know how to utilize the advanced ultra-low-power features of the SiWx917 and the Wi-Fi protocol.

This whitepaper helps you to become a master in developing energy-efficient, yet powerful and smart IoT devices using the SiWx917. It provides you with a comprehensive overview of SiWx917's low-power capabilities, power management, and Wi-Fi power saving techniques, including several best-practice tips.

Why Low-Power Wi-Fi

Historically, Wi-Fi has been a power-intensive wireless technology, prohibiting its use in energy-efficient and resource-constrained applications. However, the dramatic growth of IoT and the evolution of the energy-efficient Wi-Fi 6 protocol generation are coinciding and changing the course for the low-power Wi-Fi. With Wi-Fi 6, finally there is a simple, battery-friendly, and high-throughput way for developers to connect smart IoT devices to the cloud without bridges and protocol translations.

Wi-Fi Challenges in the IoT

Buyers seek energy-efficient, sustainable, and cost-effective products. To reflect the increasing awareness of green initiatives, new energy regulations are enforced constantly, increasing product development costs. The main challenge for the designers and developers of Wi-Fi IoT devices and smart appliances lies in power consumption.

Battery Life: Critical for IoT buyers. Long battery replacement and recharging intervals enhance user experience, reduce costs and waste, and avoid battery depletions in industrial applications, reducing OPEX.

Standby Power: Many smart devices draw 'phantom' current 24/7, increasing energy costs and CO2 emissions. Low-power Wi-Fi offers a more sustainable alternative.

Energy Regulations: Strict energy regulations are imposed constantly, increasing product development costs. Low-power Wi-Fi helps achieve better energy ratings and unlock new markets.

Where Are Low-Power Wi-Fi Devices Used?

• Smart homes: Include dozens of energy-efficient and battery-powered IoT devices like smart locks, sensors, thermostats, smart A/C units, air filters, smart switches, and energy management equipment (EV chargers, solar panel inverters, home batteries). ?
• Enterprises and Commercial Buildings: Used for sensing, controlling, access management, and building automation in commercial centers, smart hospitals, and other facilities. ?
• Smart Appliances: Subject to energy regulations, benefiting from low-power Wi-Fi connectivity. Examples include whitegoods, washers, dryers, refrigerators, countertop devices, kitchenware, BBQ grills, and fitness equipment. ?
• Location and Asset Tracking: Used in logistics, warehouses, enterprises, hospitals, and commercial centers where indoor positioning is required. ?
• Connected Health: Applications include personal health devices, physical training equipment, smart medical instruments, and asset tracking in hospitals. ❤️

SiWx917 Product Overview

Silicon Labs' SiWx917 ultra-low-power wireless SoCs and modules provide Wi-Fi 6, Bluetooth Low Energy (LE), Matter, and IP networking for secure cloud connectivity. The SiWx917 is designed for advanced battery-powered IoT devices and energy-efficient smart appliances.

The SiWx917 features a dual microprocessor architecture: a 160MHz network wireless processor (NWP) and an ARM Cortex M4 application MCU running up to 180MHz. The NWP handles TCP/IP networking, security, and wireless stacks, off-loading the MCU. The MCU is dedicated to application, Matter, and peripherals. A separate AI/ML processor offloads the MCU for optimized machine learning (ML) inferencing at the edge.

5-Year Battery Life for IoT Devices

The SiWx917 offers class-leading low power consumption, drawing only 22µA in Wi-Fi 6 connected sleep mode (without MCU, based on listed assumptions). Its combined system power consumption for Wi-Fi 6 connectivity with Target Wake Time (TWT) (connected sleep) and the application MCU is just 37µA. An independent provider, Novus Labs, estimated that the SiWx917 can enable smart locks to achieve a 5-year battery life using four AA batteries (3000mAh capacity).

SiWx917 Configurations and Assumptions for a 5-year IoT Battery-life:

• NWP in associated standby low-power mode
• SiWx917 SoC mode
• TCP client maintains socket connection
• 60-second TCP keep-alive used
• WLAN keep-alive 30 seconds
• 325kB SRAM retention
• TWT Auto Config feature enabled
• TWT Rx latency 60 sec with 8ms wakeup duration
• Arm Cortex-M4 in sleep mode (PS4) with 320kB SRAM retention
• Average current consumption for wireless and application MCU: 37µA at 3.3V
• Measurements taken in a lab
• Battery capacity: 4xAA 3000mAh

Diagram Description: A bar chart compares power consumption. The SiWx917 in Wi-Fi 6 TWT mode consumes 22µA. With Wi-Fi 6 TWT + MCU, it consumes 37µA. SiWx917 in Wi-Fi 4 mode consumes 67µA. Typical Wi-Fi solutions consume hundreds of µA.

Low-Power Wi-Fi Design Concept

Low-power Wi-Fi design can be viewed in three groups:

1. Functional Requirements: Starting from the functional requirements of the IoT application.
2. SiWx917 Ultra-Low-Power Features: Advanced features offered by the SiWx917, covering product architecture, power management, peripherals, memory, and more.
3. Wi-Fi Protocol: Optimizing power consumption through Wi-Fi protocol settings.

These three areas collectively contribute to the overall energy efficiency of an IoT device.

Diagram Description: A circular diagram illustrates the low-power Wi-Fi design concept. The outer ring represents "IoT Application Requirements". The inner rings detail "SiWx917 Ultra-Low-Power Features", including: Dual-Processor Architecture, Wi-Fi 6, Dedicated ULP SRAM, RAM Partitioning, UULP Wake-up Sources, ULP GPIO & Peripherals, Dynamic Voltage & Frequency Scaling, DTIM/TWT, Power/Clock Gating, and Power Modes & States.

SiWx917 Low-Power Features and Capabilities

This section provides an overview of the main low-power features and capabilities of the SiWx917.

Power Domains

The SiWx917 is divided into multiple power domains and partitioned subsystems for fine-grained power management using power-gating and clock-gating. This allows flexible optimization of power consumption based on IoT application requirements.

Power-Gating

Power-gating shuts off power to specific blocks when not needed, reducing consumption. Other parts can remain active to detect wake-up events. Ultra-Ultra-low-power (UULP) wake-up sources remain powered even when the Cortex-M4 MCU is in PS0 shutdown.

Examples of Power-Gating Options on SiWx917:

1. MCU Power-Gating: The CPU is power-gated in PS0 shutdown, PS1, and PS4/3/2 sleep modes, significantly reducing power. SRAM can be retained in sleep mode but not shutdown.
2. SRAM Power-Gating: A section of the 672kB SRAM can be power-gated into multiple domains for flexible management.
3. Peripheral Power-Gating: Unnecessary peripherals can be power-gated in lower power states.

Clock-Gating

Clock-gating selectively disables clock signals to blocks not in use, saving power. It allows faster wake-up compared to power-gating, which has a wake-up latency penalty. The implementation varies by operating mode and power state.

Diagram Description: A block diagram shows the SiWx917 architecture. Key blocks include: Network Wireless Processor, Power Management Unit and UULP Peripherals, Wireless Modem (Radio), Security, M4 Peripherals, Host Interface Peripherals, and AHB Interconnect. Each block contains sub-components like CPU, NWP, Cortex-M4, various peripherals (UART, I2C, GPIO, etc.), memory (SRAM, ROM, Flash), and security modules (AES, RSA). Power management features like DC-DC, LDOS, and SLEEP MANAGER are also indicated.

Power Modes and States

Cortex-M4 Application Processor

The Arm Cortex-M4 MCU offers various power modes and states as part of the SiWx917's power management strategy, enabling developers to optimize power consumption.

Active (PS4, PS3, PS2, PS1)

• PS4: Full functionality at full power.
• PS3: Functional at reduced voltage, max 90MHz MCU clock speed.
• PS2: Limited peripherals, reduced voltage, 32/20MHz MCU clock speed, SRAM retention.
• PS1: Entered from PS2. MCU power-gated, limited peripherals, RAM retention.

Sleep (PS4, PS3, PS2)

• Entered from Active states.
• MCU power-gated.
• UULP and analog peripherals available.
• Four GPIOs available as power sources (UULP_VBAT_GPIO).
• 320kB LP SRAM retained.

Standby (PS4, PS3, PS2)

• Entered from Active states.
• MCU is clock-gated.
• SRAM inherits voltage from respective Active states.
• Peripherals, GPIO, and SRAM inherit settings from respective Active states.
• Faster wake-up compared to Sleep mode (PS1, PS0).

Shutdown (PS0)

• Entered from Active states.
• MCU power-gated.
• UULP and analog peripherals available.
• Four GPIOs available as power sources (UULP_VBAT_GPIO).
• No SRAM retention.

Network Wireless Processor (NWP)

The NWP on the SiWx917 runs wireless protocols, networking stacks (TCP/IP, TLS, MQTT), and the security engine, off-loading the application MCU. It supports protocol keep-alive, allowing the MCU to sleep longer. The NWP has several power modes to optimize consumption based on network activity.

NWP power states are divided into active/high-performance and sleep/low-power states. Active states include transmit, receive, or listen. Sleep states can be connected or unconnected. Unconnected sleep can have SRAM retention or not. Connected sleep always retains RAM and wakes up based on DTIM or TWT. The NWP uses three Power Save Profiles (PSP) for data retrieval from the AP:

• Max PSP: Data retrieved via power save polling (PS-Poll). Saves power but increases delay and reduces throughput.
• Fast PSP: Streamlined data retrieval without polling. Client sends a null data frame to the AP for data retrieval.
• Enhanced Max PSP: If AP acknowledges PS-Poll but doesn't deliver data within 20ms (e.g., interoperability issue), the NWP switches to active state, sends a null data frame, and waits for Monitor Interval time.

Diagram Description: A hierarchy diagram shows NWP power management: * Active/High-performance Mode: Transmit, Receive, Listen. * Sleep/ULP Mode: With RAM Retention, Without RAM Retention. * With RAM Retention: Connected Sleep, Unconnected Sleep. * Without RAM Retention: Unconnected Sleep. * Connected Sleep can wake up based on DTIM Interval, Beacon Interval (BI), Listen Interval (LI), or Target Wake Time (TWT).

NWP Power States:

• Active: NWP is active, supports Transmit, Receive, Listen.
• Connected Sleep: NWP connects to AP and sleeps, wakes up per DTIM and TWT. SRAM Retained.
• Unconnected Sleep with Retention: NWP, modem, security engine inactive. SRAM is retained.
• Unconnected Sleep without Retention: NWP, modem, security engine inactive. SRAM is not retained.

SiWx917 State Machine

The NWP and application MCU operate based on their own state machines. The combined state diagram shows power state transitions and interactions between them.

Diagram Description: A state machine diagram illustrates the interaction between the NWP and the Cortex-M4 MCU. * NWP States: ACTIVE (works with MCU Active PS4/PS3, PS2 with internal switch), Connected Sleep (works with MCU Active, Standby, Sleep Modes; internal switch for PS2), Unconnected Sleep w/ RAM Retention (works with MCU Active, Standby, Sleep Modes), Unconnected Sleep w/o RAM Retention (works with MCU Active and Shutdown Modes). Transitions are triggered by DTIM/TWT or Wireless initialization/No wireless initiation. * MCU States: ACTIVE (PS4, PS3, PS2), STANDBY (PS4, PS3, PS2), SLEEP (PS4, PS3, PS2), SHUTDOWN (PS1, PS0). Transitions include internal switches for PS2 to work with NWP Connected Sleep, MCU power-gated in Sleep states (PS1, PS0), MCU clock-gated in Standby states. Wake-up & Reset is also shown.

Dynamic Voltage and Frequency Scaling (DVFS)

DVFS adjusts the clock frequency of the SiWx917 processor and other blocks to balance power consumption and performance. It allows switching between high-performance modes for intensive tasks and power-save modes for idle periods. The clock defaults to power-save mode after state changes.

Performance and Power-Save Modes on SiWx917

SRAM Partitioning

The SiWx917 has a total of 672kB of SRAM, segregated into different areas to reduce current consumption. It includes regular SRAM, 320kB low-power (LP) SRAM, and 8kB ultra-low-power (ULP) SRAM. These areas are further partitioned into multiple domains for flexible power management. LP SRAM and ULP SRAM can be retained in Sleep mode and PS1, but not in PS0 shutdown.

Ultra-Low-Power Peripherals

Ultra-low-power (ULP) peripherals are key to the SiWx917's low-power capabilities, enabling sensor data collection while the MCU and NWP sleep. The SiWx917 includes ULP versions of analog and digital peripherals for minimal power consumption. In Sleep mode, GPIO states are not retained. ULP pins like UULP_VBAT_GPIO can power external devices.

Peripherals with ULP Versions:

• I2C, I2S, UART, GPIO, Timers
• ADC, DAC (Analog/Digital Converters)
• DMA (Direct Memory Access)
• SSI Primary (Synchronous Serial Interface)
• RTC (Real-Time Clock)
• BOD 2 (Brown-Out Detector)

Table Description: A table shows peripheral group support across MCU Power States (Active, Standby, Sleep, Shutdown PS0) and Peripherals types (High-performance, ULP, UULP, Analog). 'X' indicates support. * Active Mode: All peripheral types are supported (PS4). * Standby Mode: All peripheral types are supported (PS4). * Sleep Mode: ULP, UULP, and Analog peripherals are supported (PS4). High-performance peripherals are not. * Shutdown PS0: ULP and UULP peripherals are supported.

DC-DC Converter

The SiWx917 includes an internal power management subsystem with DC-DC converters and linear regulators to generate required voltages from various input sources and power external components, reducing overall power consumption. Two types of DC-DC switching converters are available (1.8V or 3.3V). LC DC-DC converters power RF and digital blocks, while SC DC-DC converters power the always-on core logic.

Wi-Fi Protocol and Low-Power Design Aspects

Wi-Fi 6 is the most energy-efficient Wi-Fi generation, while Wi-Fi 4 remains ubiquitous and offers good power efficiency for many IoT applications. This section discusses significant low-power design aspects of the Wi-Fi protocol.

Delivery Traffic Indication Message (DTIM)

A DTIM is a traffic indication message where Wi-Fi APs inform clients about waiting multicast or broadcast data.

DTIM is part of the beacon message, delivered at a frequency defined by the DTIM interval (not every beacon). The DTIM interval is set by the AP in increments of 100ms for multicasting and broadcasting to multiple clients.

DTIM interval and other settings affect Wi-Fi client power consumption and battery life:

• If Listen Interval (LI) is set, the client wakes up at LI intervals to check for data. If LI is not set, the client wakes up per the AP's DTIM interval.
• The client returns to sleep after data retrieval.
• The data retrieval process, defined by the NWP's Power Save Profile (PSP), also affects power consumption.

Listen Interval (LI) can be configured on the client to wake up at multiples of the DTIM beacon interval. Setting LI to 1000ms means the client wakes up once per second (every tenth DTIM interval). An LI above 1000ms can lead to AP disconnection. TIMs in every beacon indicate pending data for a client. If the TIM bit is set, data is available for that specific client.

Diagram Description: An illustration shows device states relative to beacons. * Device Sleeps: Before a beacon with DTIM. * Device Active: Receives beacon with DTIM, sends ACK, then PS-POLL to retrieve DATA. * Device Sleeps: After data retrieval, before the next beacon without DTIM. * Device Sleeps: Before a beacon with DTIM.

Target Wake Time (TWT)

Unlike legacy modes like DTIM where Wi-Fi 4 clients wake at predefined intervals, Wi-Fi 6 TWT allows clients to negotiate wake-up schedules with the AP, preventing simultaneous wake-ups. This avoids packet collisions, reduces retransmissions, and lowers current consumption. Multicast TWT allows multiple devices to wake simultaneously. TWT enables longer sleep durations, increasing power efficiency and reducing network congestion.

Configuring individual TWT on SiWx917:

• Manual TWT Configuration: TWT parameters (sleep duration, wake-up duration, wake interval) are calculated based on TWT specification parameters in the application.
• Automatic (Auto) TWT: A Silicon Labs implementation configures TWT parameters automatically based on application requirements (average throughput, RX latency). The NWP negotiates these with the AP. Auto-TWT checks for optimal latency/throughput combinations; if inefficient, it defaults to DTIM. Auto-TWT is recommended for better throughput, interoperability, and power efficiency.
• A Wi-Fi 6 client with a TWT agreement should not transmit outside its service period (SP) to avoid media contention, improving energy efficiency.
• For frequent transmissions with short notification times (e.g., a few seconds), a short TWT interval is recommended. For sporadic low-latency transmissions (e.g., fire alarm), transmitting outside the SP might be more energy-efficient, but incurs potential media contention penalties.
• For receiving large data (e.g., firmware update), TWT teardown or re-negotiation should be considered. This adds overhead but has limited impact due to rarity.

When to Use DTIM and TWT?

While Wi-Fi 6 and TWT are recommended for low-power IoT, DTIM still has its place.

When to Use DTIM?

• Wi-Fi 6 and TWT may not be supported by all Wi-Fi routers; DTIM is generally supported.
• When network traffic is very low, the overhead of negotiating TWT is not worthwhile. DTIM requires no negotiation.
• When there is a lot of unpredictable network traffic.
• When many devices do not support TWT, to avoid collisions with TWT-enabled clients' service periods.

When to Use TWT?

• When the IoT application can tolerate long sleep times.
• When many TWT-enabled clients are on the network, and wake-up times can be arranged to avoid overlap, ensuring efficient radio resource utilization and saving power.

Conclusions – IoT-Optimized Wi-Fi

This whitepaper aimed to advance low-power Wi-Fi design. Power optimization is a complex challenge for IoT and smart device makers. Modern IoT devices require powerful compute, sophisticated features, and robust security. The SiWx917 Wi-Fi 6 wireless MCU offers a balance of ultra-low-power and advanced computing.

Benefits of SiWx917

• Maximize Battery Life ?
• Minimize Standby Current ??
• Score Better Energy Ratings ?

Features of IoT-Optimized Wi-Fi

• The lowest system power for Wi-Fi 6 and application MCU ⚡
• Dual processor for application and wireless networking ⚙️
• The largest SRAM, PSRAM, and Flash memory in the class ?
• Dedicated AI/ML processor for Edge computing ?
• Single-chip Matter solution - Pre-certified Matter and Bluetooth LE 5.4 ?
• A dedicated security engine. Broad security feature set. PSA level 2 certifiable ?️
• A dedicated security engine. Broad security feature set. PSA level 2 certifiable ?️
• Customize at order: your part number, markings, software, and other settings! ?
• A rich set of peripherals, GPIO, and Analogs with ultra-low-power operation ?

Low-Power Wi-Fi Design Guidelines for Six IoT Devices

Best-practice low-power Wi-Fi design guidelines from distinguished specialists – Six IoT device examples. Silicon Labs SiWx917 is an ultra-low-power Wi-Fi 6 wireless MCU tested to enable years of battery-life and low standby current on smart IoT devices. However, optimizing the SiWx917 wireless MCU and the Wi-Fi protocol for your design can be challenging. In this document, the distinguished specialists from the Silicon Labs Developer Services Team and Sigma Connectivity, a global design and engineering service provider, share their best-practice design guidelines for six low-power Wi-Fi device examples, helping you to avoid common pitfalls and save time.

Silicon Labs

Silicon Labs (NASDAQ: SLAB) is the leading innovator in low-power wireless connectivity, building embedded technology that connects devices and improves lives. Merging cutting-edge technology into the world's most highly integrated SoCs, Silicon Labs provides device makers with the solutions, support, and ecosystems needed to create advanced edge connectivity applications. Headquartered in Austin, Texas, Silicon Labs has operations in over 16 countries and is the trusted partner for innovative solutions in the smart home, industrial IoT, and smart cities markets. Learn more at www.silabs.com.