Computing AI at Different Power Levels
While artificial intelligence (AI) inferencing may seem seamless in daily applications, product developers and engineers on the backend know how power-hungry AI is. With each finalized product on the market came a calculated compromise to meet their energy requirements. Given how quickly AI adoption has grown over the last year1, overall energy consumption has skyrocketed, creating the need for significantly larger models and training sets. A recent study found that the AI industry could consume as much energy as a country the size of the Netherlands by 20272.
The rise of IoT has only added to the equation. Smart devices, for example, require significant energy to enable innovative use cases and lasting battery power. Efficient energy inference is an essential path forward for developers as they innovate increasingly impactful applications of AI. The bigger the network, the more you can do, right?
In this article, we’ll examine the different energy levels fueling AI compute power, their primary applications, and the future of AI inference fueled by low-power technologies.
The Energy Levels of AI Inferencing
There are three main energy levels used in AI inferencing. Each level in the hierarchy represents a trade-off between power and computational capabilities. Below, we outline those differences and how they come to life in real-world use cases:
Picojoules (pJ)
Picojoules are used for AI inferencing in ultra-low-power devices that perform simple AI tasks and face stringent power constraints. They’re also optimal for energy harvesting, the process in which energy is captured from a system’s environment and converted into usable electric power for wearables, wireless sensor networks, and more3. However, due to extremely low energy levels, picojoules may have limited computational capabilities.
Key applications in IoT at this energy level include environmental monitoring, industrial sensing, and agriculture.
Environmental monitoring involves applying AI to sensors embedded into irrigation systems, pipelines, tanks, weather stations, oceanic applications, and industrial equipment, which can detect temperature, moisture, water levels, leaks, and other physical properties. When the data is sent back to the cloud, we can track, analyze, and take further action on how resources are used and allocated.
Industrial sensing involves applying AI to temperature sensors, thermal cameras, ultrasonic sensors, photocells, inductive sensors, radar, vision cameras, and motion sensors in industrial machinery. With the sensor data, we’re able to predict and prevent machine problems before they occur.
For the agriculture industry, picojoules power levels involve applying AI to environmental sensors placed in soil or water systems. The data gathered can help with a range of issues, including crop yield prediction, selection, and management; soil compatibility classification; and water management. All aspects of smart farming contribute to feeding the growing population.
Nanojoules (nJ)
Nanojoules are often used for low-power AI applications in mobile and battery-operated devices, such as smartphones or other portable devices because they’re optimal for extending battery life and reducing the need for constant internet connection. They enable a balance between low power consumption and sufficient computational capability, so they’re suitable for a wide range of applications. While nanojoule-powered devices may be more capable than picojoule-level devices, they may still face challenges when dealing with highly complex AI computations.
Key applications in IoT at this energy level include wearables, smart homes, and remote digital health devices.
Wearables monitoring involves applying AI to gadgets — smartwatches, augmented reality technology, fitness trackers — to enhance entertainment, athletics, and other experiences. Smart home management involves applying AI to various in-home devices such as lighting, heating, speakers, etc. Machine learning picks up on the occupants’ preferences and adjusts various settings accordingly.
Remote patient monitoring is one of the fastest growing segments at this power level, which applies AI to small health wearables — pins, watches, etc. — to track patient vitals and provide personalized support, reminders about medications, and helpful insights. This use case is becoming more common as providers and patients are looking to make healthcare more efficient and result in better outcomes.
Microjoules (μJ)
Microjoules are suitable for applications where moderate power consumption is acceptable but should be minimized. They often involve more complex AI tasks, so they’re crucial for not only extending battery life but also reducing overall environmental impact. Microjoules are employed in edge computing devices where AI is done locally or in embedded systems. While still energy-efficient, microjoule-level devices may not be suitable for applications with ultra-stringent power constraints.
Key applications in IoT at this energy level include camera monitoring and automotive enhancements. Camera monitoring involves applying AI in surveillance systems, smart home cameras, motion detectors, etc. — to track and analyze humans, vehicles, attributes, and events. Automotive enhancements involve applying AI in vehicles, such as their cameras, dashboards, climate control, etc, providing new, personalized in-vehicle experiences for consumers.
How Low-Power Tech is Driving Innovation in AI Inferencing
Putting AI everywhere and incorporating it seamlessly into our daily lives is the goal. Through this transformation, power consumption will require more attention from engineers. To do more, there is a tremendous need to innovate low-power technologies that don’t impede the endpoint user’s experience. The search for the holy grail of enabling AI that consumes the least amount of power is on.
How Ambiq is Contributing
Since 2010, Ambiq, a leader of ultra low-power semiconductor solutions has helped enable over 230 million intelligent endpoint devices capable of lasting days, weeks, and even months on a single charge. Ambiq does this with the revolutionary Subthreshold Power Optimization Technology (SPOT ®) platform that provides a multi-fold in energy efficiency for smart devices. Keeping up with the increasing integration of AI within our lives, Ambiq has developed a solution that can address unique power challenges and expand the developer’s vision.
Sources:
1 The state of AI in 2023: Generative AI’s breakout year | 2023
2 Warning AI industry could use as much energy as the Netherlands | 2023
3 Three Energy-Harvesting Solutions Show Off IoT Power Possibilities | 2024
As the calendar nears the end of the first quarter of the 21st century, climate change continues to assert its worldwide presence and raise global temperatures. The culprit of extensive damage and costly recovery efforts, the annual number of floods has been on the rise for decades now1. The National Centers for Environmental Information cites flooding as the fourth most costly natural disaster, creating an average of $4.5 billion of damage per event2.
By harnessing the power of real-time data and predictive analytics, IoT-enabled sensors are revolutionizing efforts and approaches to flood detection, control, and recovery. Here, we’ll explore how IoT sensors can help minimize flood damage, highlight some real-world uses, and elucidate how this technology empowers everyone, from the government to individuals, to stay safe and dry.
Real-Time Monitoring
In the past, governments relied heavily on citizens to report flooding and prompt emergency response actions. Yet, unsurprisingly, a delayed relay of information in an emergency is far from ideal — especially in the case of a flash flood. A study published in the Ocean-Land-Atmosphere Research journal emphasized the necessity of systems that actively monitor and predict disasters are essential to manage floods and flood damage3 effectively.
The AWARE Flood Sensor from Intellisense utilizes a network of nodes installed in low-lying areas, riverbeds, and flood plains to capture real-time data via a cellular or satellite connection4. In less than a minute of startup, these nodes begin collecting and transmitting water-level data to a digital platform for users to access and review. They can warn users of flash flooding and send alerts and images to emergency responders of flood-prone areas so they can take the necessary precautions to close roads or evacuate communities.
Flood Forecasting
Traditional flood prediction and detection methods have proven inadequate in the face of increasingly severe weather patterns and rapid urbanization. However, data gathered by IoT-enabled sensors can help provide more accurate forecasts. For instance, water level and flow rate data provided by IoT sensors enables authorities to anticipate potential flooding and take preventive measures, such as reinforcing levees or activating flood barriers.
IoT-enabled sensors represent a life-changing technology in Jakarta, Indonesia, the sinking city infamous for its floods. The Jakarta government asked 2023 SAS Hackathon competitors to help develop a flood prediction system—and they got it. Hackathon team JaWaRA (Jakarta Water Resources Analytics) worked with a SAS mentor to develop an Early Warning System that’s now part of Jakarta’s citywide Flood Control System5.
The warning system uses machine learning to analyze real-time data on rainfall, water levels, and river flow to better stay ahead of flooding and mitigate it. This technology contributes to building up Jakarta’s resiliency against social, economic, and environmental disasters.
Home and Office Leak Detection
Although news channels and newspapers provide both global and area-specific flooding alerts, there are still gaps in this system that IoT-enabled sensors can help fill. For instance, flooding isn’t always due to a massive storm; sometimes, it’s due to pipes bursting or appliances malfunctioning. IoT sensors installed in buildings and homes can detect water intrusion and leakage, enabling homeowners and property managers to address issues before they escalate into costly flood damage.
Monnit develops a range of water detection solutions for homes or offices that can detect water from water heater bursts, toilet leaks, drain clogs, or any other sort of water leaks6. Their water leak detection system uses a probe point to detect changes. When water is detected, the sensor will turn on its radio and send data to a wireless gateway. This data will then be sent to the monitoring app, with alerts via email or text, enabling users to act as soon as they occur.
How Ambiq is Contributing
Using IoT-enabled sensors that relay flood-related data is becoming increasingly prevalent among individual consumers and government bodies. The IoT natural disaster detection market has a compound annual growth rate of almost 30% through the year 20307.
However, this technology will require long-lasting batteries to power the ongoing collection and transmission of data and flood-related metrics, like water levels and weather conditions. Ambiq has a range of System-on-Chips (SoCs) that can process complex data while consuming ultra-low power levels for battery-powered devices. This dramatic improvement in energy efficiency can prove vital in emergencies such as flooding that may take out power grids for days at a time. Learn more about Ambiq’s possible applications here.
Sources:
1Climate Change Indicators: Coastal Flooding | US EPA | 2023
4Commercial Flood Sensors from Intellisense – AWARE Flood IoT Sensor (intellisenseinc.com) | 2024
5SAS announces 2023 Hackathon winners | SAS | 2023
6 Best Water Leak and Flood Detection Monitoring Systems (monnit.com) | 2024
7Soaring Natural Disaster Detection IoT Market Growth (globenewswire.com) | 2024
Hearing loss affects roughly 430 million people globally1. The anguish someone feels from not being able to hear well has proven to have a startling effect on their mental health. Losing your sense of sound is difficult to adjust to, and many people feel more distanced or anxious as they no longer feel like they can contribute to conversations in social settings.
Hearing aids were developed to help facilitate sound, but they may also help improve some of the mental health suffering that comes with hearing loss. This blog will address the connection between hearing loss and anxiety, and other potential mental health issues that develop. It will detail how hearing aids can help provide users relief and look at real-world examples of hearing aids and the outcomes of adopting them.
How Hearing Aids Offers Relief
Like wearables and fitness tracking, hearing aids do more than just the basics to improve patients’ health. Hearing aids work by amplifying sound through a unique three-part system. They help improve hearing and speech comprehension of people who struggle with disabled hearing. With a microphone, amplifier, and receiver, hearing aids can assist with hearing loss caused by disease, aging, traumatic events, viral infections, and more.
While the obvious benefit of hearing aids is to help give patients the ability to hear more clearly, it can also enhance their overall well-being by making patients feel more connected to the world and how they think about themselves.
Better Relationships
Hearing loss comes with a higher risk of loneliness and social isolation, most likely due to the difficulty in understanding speech or frustration with the inability to communicate2. Hearing aids offer amplified sound so patients can hear and communicate with friends and loved ones. Overall, quality of life can improve, and patients might not feel as disconnected from others when they are able to communicate effectively.
Improved Confidence
Along with healthier relationships, hearing aids can improve confidence in communication and speech recognition. It might feel easier to accept social invitations or socialize in noisy environments. Individuals with hearing aids can experience improved confidence by appropriately communicating with the world around them.
Better Cognitive Health
Hearing loss is sometimes linked to mental and cognitive decline, so treating hearing loss through supportive hearing aids might help address that potentially harmful effect3. The reason for a potential cognitive decline could be that there is less brain stimulation and an increase in the brain working overtime to try and hear. By engaging in conversations, the person with hearing loss can keep their mind sharp and prevent this degradation.
Real-world Applications
Commercial hearing aids have been around since the turn of the 20th century. Still, innovation powered by the Internet of Things (IoT) and artificial intelligence (AI) offers many advancements over its predecessors in accessibility, performance, sound quality, and comfortable designs. Today, these are some of the leading trends in improving hearing aids.
Longer Lasting Battery Life
It’s necessary to provide reliable sound to users so they can feel confident about enjoying their day without worrying that their hearing aid will run out of batteries. The duration of your hearing aid will depend on the individual’s level of use, environmental sounds, battery size, and more.
Oticon is a rechargeable hearing aid built to provide superb sound quality and a full day of hearing4. It comes with a portable smart charger for those on the go and rely on maintaining high-quality hearing levels all day. A 30-minute charge can provide up to six hours of battery life and a full day’s charge after three hours.
Sound Adjustments Through AI
Hearing loss comes from the brain losing its ability to understand sounds processed through the cerebral cortex. Genesis AI from Starkey provides a hearing aid that enlists the help of AI to make over 80 million adjustments per hour without the need for user interference 5. Their patented Neuro Sound Technology mimics the cerebral cortex, and processes sound like a functioning auditory system does, and fills in the gaps from years of hearing loss. They offer a comfortable, ergonomic design and wireless connectivity that results in clear, high-quality sound.
Non-Traditional Design
Some of the barriers keeping people from adopting hearing aids are the design and stigma that an earpiece will affect the way others think about the individual. Along with creating more discreet earpieces that are harder to see, some manufacturers are turning to unique designs like smart glasses that act as hearing aids.
EssilorLuxottica, the parent company of popular glasses like Ray-Ban and Oakley, is developing hearing aids in the form of smart glasses to treat mild to moderate hearing loss6. These glasses feature built-in microphones and speakers that use algorithms to detect the presence of the person you are standing in front of and amplify their voice as they talk to you. Their goal is to provide a reliable and over-the-counter solution to give people back their hearing in a way that works for them.
The Future of Hearing Aids for Mental Health
As hearing loss becomes increasingly prevalent, there are promising developments in improving hearing aids to protect patients’ mental health. Advanced hearing aids could offer endpoint intelligence the potential for biometric feedback, sending real-time updates to care providers to alert them of patient activity and well-being for remote patient monitoring. Augmented reality and virtual reality technology can also be integrated for more immersive experiences, and artificial intelligence-powered personalized soundscapes could help reduce anxiety and depression.
How Ambiq Contributes
Ambiq helps next-generation endpoint devices powered by emerging technology such as AI become a reality. With Ambiq’s ultra-low power System-on-Chips (SoCs), smart battery-powered devices become more energy-efficient and operate at lower levels of power with higher performance. Visit Ambiq’s hearables applications page to learn more.
Sources:
1 Deafness and hearing loss | February 2, 2024
2 Hearing Loss, Loneliness, and Social Isolation: A Systematic Review | March 10, 2020
3 The Hidden Risks of Hearing Loss |2024
4 Oticon | 2024
5 Genesis AI | 2024
6 Essilorluxottica Nuanced Audio | 2023
The ocean is vital to the world as we rely on it for transportation, food, water, recreation, medicine, and more. However, it faces unprecedented challenges, including ecosystem destruction, overfishing, rising pollution levels, and the impacts of climate change.
In this context, leveraging advanced technologies such as artificial intelligence (AI) and the Internet of Things (IoT) becomes crucial to effectively monitor, manage, and protect marine ecosystems. The ability to collect data across various connected devices and platforms presents a powerful solution for addressing these challenges that we’ll explore in this blog.
Monitoring and Tracking Marine Species
As climate change and poor fishing practices continue to permeate our oceans, some effects can be seen in species endangerment and extinction. Fortunately, IoT-enabled tags and sensors can help monitor various marine species by tracking their movements and gathering data on migration patterns, feeding grounds, and breeding habitats. This data can then help researchers, environmentalists, and law enforcement identify areas requiring more attention and effort.
OCEARCH, a global nonprofit dedicated to accelerating the ocean’s return to balance and abundance, uses smart tags to track and monitor the sea life of more than 400 animals1. For instance, OCEARCH tracks sharks that are hunted because they are essential to a functioning ecosystem that maintains biodiversity2. Removing these apex predators from the environment could cause disruption and imbalance in the ecosystem. Users can view OCEARCH’s efforts through their app and website.
Detecting Pollutants
Every year, about 26 billion pounds (about 11,793,392,000 kg) of plastic3 and hundreds of millions of gallons of oil4 end up in our oceans. Recent IoT technology can help detect and monitor pollutants in marine environments by collecting real-time data on water quality, chemical pollutants, and oil spills.
The Liquid Robotics Wave Glider is an autonomous robot outfitted with IoT sensors5. It floats on the ocean’s surface and gathers data that was too challenging or costly to accrue in the past. The Wave Glider uses fluorometry sensors to detect substances and pollutants in the water, including crude oil and refined fuels.
As part of the research by the Philippine-California Advanced Research Institutes, they have deployed Wave Gliders to help map out the sewage discharge and polluted waters near the Boracay Islands to ensure the beaches pose no health hazards to the public6. They can then use this data to focus their cleaning efforts.
Combating Illegal Activities and Creating Safer Maritime Operations
With over 70% of the earth’s surface covered by oceans7, it’s no secret that our oceans are massive. As such, patrolling the ocean for illegal and unsafe activities is difficult. However, utilizing data-based sensors presents an innovative approach that can help reduce these activities and increase operational safety.
Pelagic Data Systems’ Vessel Tracking System supports autonomous tracking and monitoring of ocean vessels of all sizes8, enabling users to monitor ocean traffic and help ships avoid potential collisions. The Vessel Tracking System can also help users manage illegal activity in restricted-use marine areas and other activities that pose a significant threat to marine conservation and sustainability.
Outlook of Technology in Maritime Operations
Looking ahead, the future of connected sensors in maritime operations is promising. Advances in sensor technology, data analytics, and connectivity will continue to drive innovation in marine conservation and safety. From autonomous underwater vehicles to satellite-based monitoring systems, IoT and AI will play an increasingly pivotal role in protecting our oceans, preserving marine biodiversity, and ensuring sustainable maritime practices.
As technology advances, continued investment in these solutions is critical to safeguarding our marine ecosystems for future generations.
How Ambiq Is Contributing
AI and IoT are powerful tools in marine conservation and safety, providing a foundation for impactful solutions that can address our oceans’ conservation and safety challenges.
That said, there aren’t any power outlets in the middle of the ocean. Consequently, marine IoT technology requires long-lasting batteries to power the ongoing collection and transmission of ocean-related data. Ambiq’s System-on-Chips (SoCs) both enhance data processing capabilities and extend battery life while consuming ultra-low power levels. Discover and learn more about Ambiq’s possible applications here.
Sources:
1 About – OCEARCH | 2024
3 Ocean pollution – 11 facts you need to know (conservation.org)
Four https://oceantoday.noaa.gov/oilintheocean/
5 The Wave Glider | Unmanned Surface Vehicle by Liquid Robotics (liquid-robotics.com)
6 Monitoring Reefs, Pollution, and Mariculture in the Philippines | March 25, 2019
7 How much water is in the ocean? (noaa.gov)
It’s safe to call the Internet of Things (IoT) a fact of life with a projection that there will be around 17 billion devices in 20241. And unlike some sectors, exponential growth isn’t leading to rapid stagnation. IoT is constantly advancing, and ground-shaking innovations are par for the course.
2024 is already shaping up to deliver major transformations, and big things are happening at the broader societal level. Here are four trends to look out for in the days to come:
Architectures Become More Flexible
As technology evolves and becomes more diverse, the demand to create more cohesive networks that developers can seamlessly build across various platforms will become more prevalent.
Open-sourced architecture has become a winning solution. By allowing collaboration, improvements, and flexibility, developers can use open-source architectures to adapt to their application’s specific needs. This results in agile environments, higher scalability, and continuous improvements.
Recently, Xiaomi announced the Vela Platform at their IoT Ecological partner conference. The Vela platform is an embedded software platform built on the open-source real-time operating system NuttX. Vela provides unified software services on various IoT hardware platforms, supporting multiple components and a user-friendly framework to integrate fragmented IoT application scenarios.
It has a range of tools that help developers debug and is already picking up support from big names in integrated circuits including Ambiq with our Apollo4 Blue Display Kit designed to show off the graphical and technical capabilities of our system on chips (SoCs) for IoT devices.
Ultra-low Power Consumption Will Start to Become the Norm
AI is becoming increasingly integral to business workflows. However, depending on an external cloud isn’t always viable when your sensors and other assets are on-premises. If your sector leans heavily on IoT devices, you’ll have to rethink how you use power to keep up.
Artificial Intelligence (AI) is known for consuming massive amounts of power. While the training phase commonly shoulders most of the blame, studies show that inference is also energy-hungry2. This doesn’t bode well for use cases like stock management, early disease diagnosis, and other innovative applications.
Ultra-low-power SoCs will take center stage in the push for IoT-powered AI sustainability. The increasing availability of AI-enabled endpoint devices will only spur demand. Low-voltage technology like Ambiq’s Subthreshold Power Optimized Technology (SPOT®) has proven that high-end computing is possible without massive energy consumption. Now, it’s just a matter of putting the pieces together.
IoT Cybersecurity Will Blossom for Endpoint Devices
AI is a powerful tool, so what happens when it winds up in the wrong hands? Unfortunately, we’re already starting to see the ramifications play out. AI-generated hacks are emerging among the biggest cybersecurity threats of the modern era.
AI can make finding exploits far less time-consuming and effective. Worse, the wide accessibility of cloud technology makes it all the more appealing to cyber criminals3. This is already happening in the consumer sector, with potentially devastating results. Email security provider Egress revealed that AI makes phishing campaigns harder to detect4. It’s also helping bad actors sidestep quarantines and detectors.
The good news is that fighting fire with fire could be a valid defense. ML models like support vector machines and random forests have been shown to detect suspicious activity with high accuracy and efficient memory usage5. If your most vulnerable assets become compromised, embedding AI-capable hardware right on the endpoint could be an effective countermeasure to regain control over those assets or minimize the damage.
Regulations May Grow Sharper Teeth
While not an exciting trend, IoT regulatory compliance is becoming more of a necessity by the second. Governments may be warming up to IoT like never before, but they haven’t thrown all caution to the wind.
In early December 2023, the US Office of Management and Budget (OMB) exemplified this in a new guidance. To comply with the OMB’s rules, federal agencies must provide detailed inventories of their IoT devices by the end of fiscal year 20246.
The US wasn’t the only nation to move in this direction. In October, the UK government finalized a sweeping set of security regulations for connectable products — devices that either connect to the internet or send and receive data via EM transmissions7. These rules targeted manufacturers, covering everything from password strength to updates.
It wouldn’t be surprising to see vendors face similar requirements. Also, data privacy laws like the EU’s GDPR have long since established that being a foreign-domiciled enterprise doesn’t exempt you from compliance burdens. Regulatory crackdowns don’t necessarily mean application growth has to slow; it just needs a more comprehensive approach to governance. Hardware vendors can and should play a role in this journey.
How Ambiq is Contributing
These predictions aren’t exclusive to 2024. They’ll continue to pose concerns in the years to follow — and waiting to play catch up isn’t quite a winning strategy.
Ambiq’s ultra-low power portfolio empowers you to meet increased processing demand and adapt to the changing landscape. From driving Arm® Cortex® M4 energy consumption lower than thought possible with the Apollo4 Blue Plus SoC to ensuring always-on devices live up to their name. We’re helping IoT reach its true potential. Dive into our catalog to find out more.
Sources:
1 Internet of Things (IoT) in the US – statistics & facts | December 19, 2023
2 OMB guidance asks agencies to provide inventory of IoT assets | December 6, 2023
3 Prepare for AI Hackers | March-April 2023
4 2023 Phishing Threat Trends Report | 2023
5 Detecting Cybersecurity Attacks in Internet of Things Using Artificial Intelligence Methods: A Systematic Literature Review | January 10, 2022
6 The growing energy footprint of artificial intelligence | October 18, 2023
7 UK government finalises IoT cybersecurity requirements | October 25, 2023