The New Skills Behind Smart Sensor Innovation
Miniaturization, AI, and Battery-Free IoT
Image Source: rawdesign/stock.adobe.com; generated with AI
By Brandon Lewis for Mouser Electronics
Published October 3, 2025
Smart sensing is radically changing how organizations collect and analyze data to better understand environments, processes, and even patterns of human behavior.[1] Building intelligence into sensor systems enables more accurate and automated data collection, minimizing the need for 24/7 human monitoring. It allows routine processes to be automated more effectively, lowering operational costs and improving building systems, industrial safety, and farming workflows.
A typical smart sensor system uses an optimized combination of sensing components, signal chain circuitry like amplifiers, and computing elements such as microcontroller units (MCUs). However, recent progress in sensor miniaturization, integrated functions, and edge artificial intelligence (edge AI) has changed how engineers approach smart sensor system design. This article examines the impact of sensor component evolution on design processes. It explores primary smart sensor industry trends, including ambient Internet of Things (IoT)- and AI-enabled MCUs and the changing expertise required to address new design challenges.
Modern Sensors as Integrated Subsystems
Two breakthroughs have resulted in a move away from bulky and fragile sensors to today’s compact, intelligent devices.
- Micro-electromechanical systems (MEMS) technology has made it possible to shrink sensors to microscopic sizes, and improvements in MEMS manufacturing have increased their reliability and precision.[2]
- Digitization removes the many challenges of analog sensing by reducing the external signal chains required to obtain reliable, accurate data.
As a result, many modern sensors are ready-to-go solutions that require less component-level design expertise. This enables faster product development while reducing design costs, the number of signal chain components, and calibration requirements.
Improvements in semiconductor manufacturing have enabled sensors to integrate features like power-saving modes, wake-up triggers, and built-in AI, which improve efficiency or pre-process data. With these capabilities, sensors can act more like modular subsystems with logic and autonomy, instead of simply being digital data sources. Even with these added functions, modern sensors remain highly compact, promoting smaller smart devices that consume less power than earlier designs.
As sensors gain more built-in functions, design is shifting away from component-level work toward application- and data-level design. The new challenge is ensuring that software and firmware remain compatible as these devices become more complex.
Low-Power Sensing and the Rise of Ambient IoT
The decreasing size, power consumption, and cost of sensors are key factors in the widespread adoption of IoT across factories, supply chains, buildings, and infrastructure. Wireless sensing modules can monitor many physical variables such as temperature, vibration, pressure, gases, and motion in previously unworkable locations.
However, wireless IoT sensors are typically battery-powered, introducing maintenance and scalability challenges. For example, regularly replacing or recharging batteries in a factory that uses thousands of sensors becomes expensive, time-consuming, and logistically complex. As a result, IoT sensor nodes must consume extremely low power to prolong operating life and reduce maintenance and downtime. While power benchmarks are essential, this ultra-low-power operation depends more on sleep mode optimization in firmware than on hardware choices.
As sensor sleep currents have dropped to the microamp level, designers can now combine them with ambient IoT, a new technology that aims to remove disposable batteries from designs altogether. This new technology, which comprises self-powered, ultra-low-energy endpoints, proposes that many wireless sensor nodes could operate solely on harvested micro-renewable energies like kinetic, RF, thermal, or ambient light. By leveraging emerging energy-harvesting technologies, devices can achieve battery-free operation and achieve significant benefits:
- Sustainability: Harvesting ambient energy and aligning with environmental sustainability goals eliminates millions of disposable batteries, which are estimated to be consumed at a rate of 78 million per day globally.[3]
- Scalability: Ambient energy removes the need for battery replacement, enabling economically viable deployment of thousands of sensors.
- Set-and-forget operation: Sensors can be installed in hard-to-reach or embedded locations without ongoing maintenance.
ABI Research estimates that 1.1 billion ambient IoT devices will be shipped by 2030.[4] This growth drives new opportunities where sensor hardware design requires new areas of expertise:
- Selecting and operating sensors and wireless MCUs with microamp- or nanoamp-level sleep currents in strict power constraints
- Designing for efficient component interoperability to maximize device performance
- Understanding energy-harvesting power infrastructure, including:
- Storage elements like supercapacitors or rechargeable batteries
- Energy harvesters like photovoltaic cells or piezoelectric transducers
- Power management integrated circuits (PMICs) for DC-DC conversion and energy distribution
While ambient IoT demands software expertise similar to IoT development, ensuring installed devices can run indefinitely relies on hardware design expertise that few developers possess for such narrow power envelopes.
Software developers interested in the smart element of ambient IoT systems usually achieve this intelligence in the cloud due to the hardware power constraints.
The Value of AI Inference with Raw Sensor Data
Besides real-time tasks like security and industrial process monitoring, sensor-driven IoT at the edge typically reports data to the cloud periodically instead of continuously. This is different from the kilohertz-level sampling capabilities of modern sensors. While high volumes of data are not necessarily useful, standard IoT does not unlock the full potential of these devices.
AI-ready MCUs allow sensor devices to process fast data streams at the edge rather than sending every data point to the cloud. AI acts as a filter to turn a constant data stream into high-value event triggers and inferences, without the raw data leaving the device. This technique enhances data security and enables sensors to make decisions independently without always relying on a network connection.
On-device AI is also changing the way smart sensors are designed. Instead of waking a system whenever a simple threshold is crossed, AI allows the device to wake only when a specific pattern or event is detected in the sensor data. To enable this, engineers need to train and test AI models so that the system can recognize those events accurately. The most reliable triggers often come from combining data from multiple sensors rather than using a single sensor in isolation. This approach is known as sensor-fusion AI and is particularly useful when the target behavior can only be detected by combining data from multiple sensors.[5]
For example, doctors may combine a wearable heart rate monitor with accelerometers and temperature sensors to monitor the condition of a cardiac patient. Through sensor-fusion AI, the wearable monitor can detect anomalous heart activity that, when combined with specific body movements or temperature changes, may indicate a mild episode, versus a more severe episode that suggests the patient is in danger. By performing AI processing at the edge and sending inference notifications only in case of emergency, the monitor allows the patient to remain independent during recovery with the reassurance that their health data are secure.
As with today’s advanced sensors, bringing AI to the edge means developers must build new design and development skills. When running AI on edge devices like MCUs, developers can use tools such as Edge Impulse and TensorFlow Lite to optimize models, handle parameter quantization to fit within tight memory limits, and simplify deployment on resource-constrained hardware.
New Paths for Smart Sensor Developers
The fast-moving advancements in IoT sensor technology, from size to power to AI capabilities, create new opportunities for businesses and the development community. To deliver on the promise of smart sensors, new proficiency levels are required. In many application areas, the developer’s role is moving from low-level hardware optimization to system-level design and software engineering.
With smart sensing systems defined more by integration than by individual parts, success now depends on balancing hardware, software, data expertise, and staying current with modern industry trends.
Ambient IoT opens new possibilities for hardware developers, requiring them to leverage skills in low-power design and energy harvesting. Edge AI is doing the same for software developers as they become data science and model engineering experts. Together, these trends change how smart sensor systems work. They’re likely to converge as sensor technology continues to advance.
Sources
[1]https://www.techtarget.com/iotagenda/definition/smart-sensor
[2]https://esenssys.com/news/comprehensive-guide-to-mems-sensors/
[3]https://www.europeandissemination.eu/up-to-78-million-batteries-will-be-discarded-daily-by-2025-researchers-warn/14495
[4]https://www.rcrwireless.com/20250425/internet-of-things/billion-ambient-iot
[5]https://www.nature.com/articles/s41746-023-00897-6