\n Media contact:<\/strong> Kayla Albert, 765-494-2432, wiles5@purdue.edu<\/a> Note to journalists:<\/p>\n \n For a copy of a research paper mentioned in this story, please contact Kayla Albert at wiles5@purdue.edu<\/a> or 765-494-2432. High-resolution versions of the story\u2019s photos<\/a> and b-roll of Purdue University\u2019s campus<\/a> are available via Google Drive. <\/p>\n <\/div>\n <\/div>\n <\/div>\n<\/div>\n","protected":false},"excerpt":{"rendered":" WEST LAFAYETTE, Ind. \u2014 Is it possible to use artificial intelligence tools like ChatGPT without internet access? Not yet, but if it were possible, it wouldn\u2019t just expand what we could do with those tools \u2014 it also might save<\/p>\n","protected":false},"author":25,"featured_media":8467,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[7],"tags":[],"department":[],"source":[29],"purdue_today_topic":[],"coauthors":[131],"class_list":["post-8774","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-research-excellence","source-purdue-news"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/posts\/8774","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/users\/25"}],"replies":[{"embeddable":true,"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/comments?post=8774"}],"version-history":[{"count":2,"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/posts\/8774\/revisions"}],"predecessor-version":[{"id":8776,"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/posts\/8774\/revisions\/8776"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/media\/8467"}],"wp:attachment":[{"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/media?parent=8774"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/categories?post=8774"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/tags?post=8774"},{"taxonomy":"department","embeddable":true,"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/department?post=8774"},{"taxonomy":"source","embeddable":true,"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/source?post=8774"},{"taxonomy":"purdue_today_topic","embeddable":true,"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/purdue_today_topic?post=8774"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/www.purdue.edu\/newsroom\/wp-json\/wp\/v2\/coauthors?post=8774"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
\u00a0
Source:<\/strong> Shreyas Sen, shreyas@purdue.edu<\/a> <\/p>\n <\/div>\n
additional information<\/span><\/h2>\n \n <\/div>\n\n \n - \n \n Now and then, on and off campus: Semiconductors at Purdue <\/a>\n <\/li>\n
- \n \n Latest in AI research: Improving life and wellness through innovation <\/a>\n <\/li>\n
- \n \n Purdue Computes <\/a>\n <\/li>\n <\/ul>\n\n<\/div>\n\n\n\n\n
Currently, AI\u2019s large power consumption is a major reason why it is only accessible using the internet. AI algorithms are constrained to data centers \u2014 massive facilities with servers that make up the internet. Devices such as your smartphone or smartwatch access AI algorithms from data centers over the cloud.<\/p>\n\n\n\n
The tech industry is working toward creating \u201cedge\u201d technology, which is a chip that would run AI algorithms directly inside of smartphones or other devices instead of having to access these algorithms via the cloud.<\/p>\n\n\n\n
But doing something as complex as learning the whole internet, like ChatGPT does, isn\u2019t possible for a chip right now because it requires too much energy. Sen is custom-designing chips for wearable devices that would allow them to work with edge devices to handle more complex problems with AI using less power, similarly to how a human\u2019s nerves and brain work together.<\/p>\n\n\n\n![\"A](\"https:\/\/www.purdue.edu\/newsroom\/wp-content\/uploads\/2024\/08\/sen-chipcloseup.jpg\" "\\"\\"")
This patented integrated circuit invented by Shreyas Sen and his students was among the first of their chip designs inspired by the human body. (Purdue University photo\/Kelsey Lefever)<\/figcaption><\/figure>\n\n\n\n\u201cIf we want to significantly increase the battery life or functionality of a wearable AI device, then we need new mathematical or physical concepts implemented into custom circuit designs. These are not components we can just buy off the market,\u201d Sen said.<\/p>\n\n\n\n
How the human body could inform chip design<\/h2>\n\n\n\n
Custom-designed chips that incorporate the latest advances in mathematical and physical concepts are Purdue\u2019s specialty.<\/p>\n\n\n\n
Each year, researchers from all over the world present integrated circuit designs that have the greatest potential to outperform the tech industry\u2019s state-of-the-art chips at the IEEE International Solid-State Circuits Conference (ISSCC)<\/a>. Purdue is emerging as one of the top universities whose chip designs are selected for presentation at this conference and publication in the IEEE Journal of Solid-State Circuits<\/a>. Within the past 25 years, Purdue researchers have published more than 30 papers<\/a> in this journal, which is considered the flagship journal in chip design.<\/p>\n\n\n\nIn recent years, 10 of those Purdue papers described chip designs from Sen\u2019s lab. These chips include ones designed for transferring data at high speeds and ultra-low power levels, which are needed for using AI directly on wearable devices such as smartwatches.<\/p>\n\n\n\n
This energy efficiency is possible because these chips allow wearable devices to create a network that mimics the human body\u2019s nervous system.<\/p>\n\n\n\n![\"An](\"https:\/\/www.purdue.edu\/newsroom\/wp-content\/uploads\/2024\/08\/sen-chipcorner.jpg\" "\\"\\"")
Shreyas Sen and his students regularly publish in journals and attend conferences key to the field of integrated circuit design. This patent-pending chip design, presented at the IEEE International Solid-State Circuits Conference in 2022, may help a network of wearable devices to transfer video to each other faster using less power. (Purdue University photo\/Kelsey Lefever)<\/figcaption><\/figure>\n\n\n\nA human\u2019s nervous system has two main parts: the central nervous system, which includes the brain, and the peripheral nervous system, which has nerves that the brain uses to rapidly transmit information throughout the body. Even though the brain consumes more energy than most organs of the body, it offsets this energy use<\/a> by maximizing the amount of information it transmits per unit of energy via the peripheral nervous system.<\/p>\n\n\n\nThe wearable devices we use to access the internet, including AI algorithms over the cloud, have chips with \u201cbrains\u201d in the form of central processing units, but no \u201cnerves.\u201d Without having something to function as a peripheral nervous system, these \u201cbrains\u201d in wearable devices use high-energy electromagnetic waves such as Bluetooth or Wi-Fi to send information to each other, which limits the battery life of these devices and prevents them from having the capability to support complex AI algorithms on-chip.<\/p>\n\n\n\n
The chip designs Sen\u2019s lab has developed supply an artificial peripheral nervous system by transmitting information over lower-frequency signals. Using these signals, which are in the electro-quasistatic range on the electromagnetic spectrum, these chips can send data to each other 10 times faster than Bluetooth while consuming 100 times less energy per bit than Bluetooth or Wi-Fi<\/a>. At ISSCC in 2022, Sen and his students presented a patent-pending chip design that uses electro-quasistatic signals to address problems with increasing the speed and energy efficiency of transferring video between wearable devices<\/a>.<\/p>\n\n\n\n![\"A](\"https:\/\/www.purdue.edu\/newsroom\/wp-content\/uploads\/2024\/08\/sen-chipchart.jpg\" "\\"\\"")
Each row of small rectangles in this photo displays a generation of a patent-pending chip design that beats Bluetooth data transfer speeds while consuming less energy per bit. With each generation, Shreyas Sen has been developing this design into a commercial product called \u201cWi-R\u201d through Ixana, a startup he cofounded. (Purdue University photo\/Kelsey Lefever)<\/figcaption><\/figure>\n\n\n\nToward designing chips that use AI without needing the whole internet<\/h2>\n\n\n\n
Sen and his collaborators have begun testing out how to incorporate AI algorithms into this high-speed, low-power chip design concept for wearable devices.<\/p>\n\n\n\n
Among his papers published in the IEEE Journal of Solid-State Circuits, Sen has demonstrated patent-pending state-of-the-art chips that use AI algorithms in a method called \u201cin-sensor analytics\u201d<\/a> to allow a wearable device to only interpret the data it needs to perform certain functions.<\/p>\n\n\n\nLet\u2019s say you have an app on your smartphone that uses an electrocardiogram patch with an in-sensor analytics chip on your chest to notify you when you\u2019re experiencing irregular heart activity, regardless of a good internet connection.<\/p>\n\n\n\n
Even though the smartphone would have internet access as an edge device, the in-sensor analytics chip within the patch wouldn\u2019t need constant internet access to use AI because these algorithms would have already been trained to only interpret irregular heart activity.<\/p>\n\n\n\n
In addition to not having to worry about internet access, you also wouldn\u2019t have to stress over keeping your wearable electrocardiogram charged thanks to a low-powered \u201cperipheral nervous system\u201d \u2014 in the form of electro-quasistatic signals \u2014 that connects your patch to your smartphone.<\/p>\n\n\n\n
Sen envisions that other wearable devices ranging from smartwatches and smart glasses to implants could also tap into this network to do more with AI if equipped to transfer data using electro-quasistatic signals. Last year, Sen\u2019s lab published initial results<\/a> for a patent-pending implant concept it developed that was the first to demonstrate electro-quasistatic signal communication in the brain.<\/p>\n\n\n\n![\"An](\"https:\/\/www.purdue.edu\/newsroom\/wp-content\/uploads\/2024\/08\/sen-chiparrayOG-1024x538.jpg\" "\\"\\"")
This array shows a patent-pending chip for a brain implant concept Shreyas Sen\u2019s lab is developing. Sen\u2019s goal is for wearable devices and implants to be part of a network of electro-quasistatic signals that supports complex AI tasks using less power. (Purdue University photo\/Kelsey Lefever)<\/figcaption><\/figure>\n\n\n\nAlthough Sen hasn\u2019t yet designed a chip for creating an AI-driven electrocardiogram patch, he is working on bringing to market a chip that might make that kind of application possible someday. In 2020, Sen, his students and other Purdue alumni founded a startup called Ixana<\/a> to commercialize a patent-pending chip, \u201cWi-R,\u201d<\/a> that incorporates their designs inspired by the human body\u2019s nervous system.<\/p>\n\n\n\nSen and his team debuted Wi-R last year at CES<\/a>, an annual technology show in Las Vegas. This year, the chip was named an honoree of two CES innovation awards<\/a>. Ixana was also recently included in the Electronic Engineering Times\u2019 annual list of 100 startups<\/a> in the silicon industry to watch in 2024.<\/p>\n\n\n\nThrough this startup and his continued published research, Sen is helping the tech industry figure out how to incorporate lessons from the human body into wearable AI devices.<\/p>\n\n\n\n
\u201cWith custom-made chips, we could overcome the energy constraints of wearable devices that prevent them from using AI to solve increasingly complex problems,\u201d Sen said. \u201cThis could lead to the development of devices that we aren\u2019t able to imagine yet.\u201d<\/p>\n\n\n\n![\"A](\"https:\/\/www.purdue.edu\/newsroom\/wp-content\/uploads\/2024\/08\/sen-tinychip.jpg\" "\\"\\"")
Ixana\u2019s \u201cWi-R\u201d chip, born out of Shreyas Sen\u2019s lab, has received awards at CES, an annual technology show in Las Vegas. (Purdue University photo\/Kelsey Lefever)<\/figcaption><\/figure>\n\n\n\nThe Purdue Innovates Office of Technology Commercialization<\/a> has filed patent applications wherever applicable for Sen\u2019s chip designs. Sen\u2019s work in chip design has been partly funded by the National Science Foundation and more than 20 companies from multiple industries. The Elmore Family School of Electrical and Computer Engineering is one of Purdue\u2019s computing departments, which are part of the Purdue Computes<\/a> initiative. Sen also is a researcher in the Scalable Manufacturing of Aware & Responsive Thin Films Consortium<\/a> at Purdue. This consortium is affiliated with the university\u2019s Institute for Physical Artificial Intelligence<\/a>.<\/p>\n\n\n\nAbout Purdue University<\/h2>\n\n\n\n
Purdue University is a public research institution demonstrating excellence at scale. Ranked among top 10 public universities and with two colleges in the top four in the United States, Purdue discovers and disseminates knowledge with a quality and at a scale second to none. More than 105,000 students study at Purdue across modalities and locations, including nearly 50,000 in person on the West Lafayette campus. Committed to affordability and accessibility, Purdue\u2019s main campus has frozen tuition 13 years in a row. See how Purdue never stops in the persistent pursuit of the next giant leap \u2014 including its first comprehensive urban campus in Indianapolis, the\u202fMitch Daniels School of Business, Purdue Computes and the One Health initiative\u202f\u2014 at\u202fhttps:\/\/www.purdue.edu\/president\/strategic-initiatives<\/a>.<\/p>\n\n\n\nPapers<\/h2>\n\n\n\n
Invited: Human-inspired distributed wearable AI<\/em>
2024 Design Automation Conference
DOI: 10.48550\/arXiv.2406.18791<\/a><\/p>\n\n\n\nBodyWire: A 6.3-pJ\/b 30-Mb\/s \u221230-dB SIR-tolerant broadband interference-robust human body communication transceiver using time domain interference rejection<\/em>
IEEE Journal of Solid-State Circuits
DOI: 10.1109\/JSSC.2019.2932852<\/a><\/p>\n\n\n\nA 65nm 63.3\u00b5W 15Mbps transceiver with switched-capacitor adiabatic signaling and combinatorial-pulse-position modulation for body-worn video-sensing AR nodes<\/em>
2022 IEEE International Solid-State Circuits Conference
DOI: 10.1109\/ISSCC42614.2022.9731793<\/a><\/p>\n\n\n\nA 65 nm wireless image SoC supporting on-chip DNN optimization and real-time computation-communication trade-off via actor-critical neuro-controller<\/em>
IEEE Journal of Solid-State Circuits
DOI: 10.1109\/JSSC.2022.3159473<\/a><\/p>\n\n\n\n
Currently, AI\u2019s large power consumption is a major reason why it is only accessible using the internet. AI algorithms are constrained to data centers \u2014 massive facilities with servers that make up the internet. Devices such as your smartphone or smartwatch access AI algorithms from data centers over the cloud.<\/p>\n\n\n\n
The tech industry is working toward creating \u201cedge\u201d technology, which is a chip that would run AI algorithms directly inside of smartphones or other devices instead of having to access these algorithms via the cloud.<\/p>\n\n\n\n
But doing something as complex as learning the whole internet, like ChatGPT does, isn\u2019t possible for a chip right now because it requires too much energy. Sen is custom-designing chips for wearable devices that would allow them to work with edge devices to handle more complex problems with AI using less power, similarly to how a human\u2019s nerves and brain work together.<\/p>\n\n\n\n \u201cIf we want to significantly increase the battery life or functionality of a wearable AI device, then we need new mathematical or physical concepts implemented into custom circuit designs. These are not components we can just buy off the market,\u201d Sen said.<\/p>\n\n\n\n Custom-designed chips that incorporate the latest advances in mathematical and physical concepts are Purdue\u2019s specialty.<\/p>\n\n\n\n Each year, researchers from all over the world present integrated circuit designs that have the greatest potential to outperform the tech industry\u2019s state-of-the-art chips at the IEEE International Solid-State Circuits Conference (ISSCC)<\/a>. Purdue is emerging as one of the top universities whose chip designs are selected for presentation at this conference and publication in the IEEE Journal of Solid-State Circuits<\/a>. Within the past 25 years, Purdue researchers have published more than 30 papers<\/a> in this journal, which is considered the flagship journal in chip design.<\/p>\n\n\n\n In recent years, 10 of those Purdue papers described chip designs from Sen\u2019s lab. These chips include ones designed for transferring data at high speeds and ultra-low power levels, which are needed for using AI directly on wearable devices such as smartwatches.<\/p>\n\n\n\n This energy efficiency is possible because these chips allow wearable devices to create a network that mimics the human body\u2019s nervous system.<\/p>\n\n\n\n A human\u2019s nervous system has two main parts: the central nervous system, which includes the brain, and the peripheral nervous system, which has nerves that the brain uses to rapidly transmit information throughout the body. Even though the brain consumes more energy than most organs of the body, it offsets this energy use<\/a> by maximizing the amount of information it transmits per unit of energy via the peripheral nervous system.<\/p>\n\n\n\n The wearable devices we use to access the internet, including AI algorithms over the cloud, have chips with \u201cbrains\u201d in the form of central processing units, but no \u201cnerves.\u201d Without having something to function as a peripheral nervous system, these \u201cbrains\u201d in wearable devices use high-energy electromagnetic waves such as Bluetooth or Wi-Fi to send information to each other, which limits the battery life of these devices and prevents them from having the capability to support complex AI algorithms on-chip.<\/p>\n\n\n\n The chip designs Sen\u2019s lab has developed supply an artificial peripheral nervous system by transmitting information over lower-frequency signals. Using these signals, which are in the electro-quasistatic range on the electromagnetic spectrum, these chips can send data to each other 10 times faster than Bluetooth while consuming 100 times less energy per bit than Bluetooth or Wi-Fi<\/a>. At ISSCC in 2022, Sen and his students presented a patent-pending chip design that uses electro-quasistatic signals to address problems with increasing the speed and energy efficiency of transferring video between wearable devices<\/a>.<\/p>\n\n\n\n Sen and his collaborators have begun testing out how to incorporate AI algorithms into this high-speed, low-power chip design concept for wearable devices.<\/p>\n\n\n\n Among his papers published in the IEEE Journal of Solid-State Circuits, Sen has demonstrated patent-pending state-of-the-art chips that use AI algorithms in a method called \u201cin-sensor analytics\u201d<\/a> to allow a wearable device to only interpret the data it needs to perform certain functions.<\/p>\n\n\n\n Let\u2019s say you have an app on your smartphone that uses an electrocardiogram patch with an in-sensor analytics chip on your chest to notify you when you\u2019re experiencing irregular heart activity, regardless of a good internet connection.<\/p>\n\n\n\n Even though the smartphone would have internet access as an edge device, the in-sensor analytics chip within the patch wouldn\u2019t need constant internet access to use AI because these algorithms would have already been trained to only interpret irregular heart activity.<\/p>\n\n\n\n In addition to not having to worry about internet access, you also wouldn\u2019t have to stress over keeping your wearable electrocardiogram charged thanks to a low-powered \u201cperipheral nervous system\u201d \u2014 in the form of electro-quasistatic signals \u2014 that connects your patch to your smartphone.<\/p>\n\n\n\n Sen envisions that other wearable devices ranging from smartwatches and smart glasses to implants could also tap into this network to do more with AI if equipped to transfer data using electro-quasistatic signals. Last year, Sen\u2019s lab published initial results<\/a> for a patent-pending implant concept it developed that was the first to demonstrate electro-quasistatic signal communication in the brain.<\/p>\n\n\n\n Although Sen hasn\u2019t yet designed a chip for creating an AI-driven electrocardiogram patch, he is working on bringing to market a chip that might make that kind of application possible someday. In 2020, Sen, his students and other Purdue alumni founded a startup called Ixana<\/a> to commercialize a patent-pending chip, \u201cWi-R,\u201d<\/a> that incorporates their designs inspired by the human body\u2019s nervous system.<\/p>\n\n\n\n Sen and his team debuted Wi-R last year at CES<\/a>, an annual technology show in Las Vegas. This year, the chip was named an honoree of two CES innovation awards<\/a>. Ixana was also recently included in the Electronic Engineering Times\u2019 annual list of 100 startups<\/a> in the silicon industry to watch in 2024.<\/p>\n\n\n\n Through this startup and his continued published research, Sen is helping the tech industry figure out how to incorporate lessons from the human body into wearable AI devices.<\/p>\n\n\n\n \u201cWith custom-made chips, we could overcome the energy constraints of wearable devices that prevent them from using AI to solve increasingly complex problems,\u201d Sen said. \u201cThis could lead to the development of devices that we aren\u2019t able to imagine yet.\u201d<\/p>\n\n\n\n The Purdue Innovates Office of Technology Commercialization<\/a> has filed patent applications wherever applicable for Sen\u2019s chip designs. Sen\u2019s work in chip design has been partly funded by the National Science Foundation and more than 20 companies from multiple industries. The Elmore Family School of Electrical and Computer Engineering is one of Purdue\u2019s computing departments, which are part of the Purdue Computes<\/a> initiative. Sen also is a researcher in the Scalable Manufacturing of Aware & Responsive Thin Films Consortium<\/a> at Purdue. This consortium is affiliated with the university\u2019s Institute for Physical Artificial Intelligence<\/a>.<\/p>\n\n\n\n Purdue University is a public research institution demonstrating excellence at scale. Ranked among top 10 public universities and with two colleges in the top four in the United States, Purdue discovers and disseminates knowledge with a quality and at a scale second to none. More than 105,000 students study at Purdue across modalities and locations, including nearly 50,000 in person on the West Lafayette campus. Committed to affordability and accessibility, Purdue\u2019s main campus has frozen tuition 13 years in a row. See how Purdue never stops in the persistent pursuit of the next giant leap \u2014 including its first comprehensive urban campus in Indianapolis, the\u202fMitch Daniels School of Business, Purdue Computes and the One Health initiative\u202f\u2014 at\u202fhttps:\/\/www.purdue.edu\/president\/strategic-initiatives<\/a>.<\/p>\n\n\n\n Invited: Human-inspired distributed wearable AI<\/em> BodyWire: A 6.3-pJ\/b 30-Mb\/s \u221230-dB SIR-tolerant broadband interference-robust human body communication transceiver using time domain interference rejection<\/em> A 65nm 63.3\u00b5W 15Mbps transceiver with switched-capacitor adiabatic signaling and combinatorial-pulse-position modulation for body-worn video-sensing AR nodes<\/em> A 65 nm wireless image SoC supporting on-chip DNN optimization and real-time computation-communication trade-off via actor-critical neuro-controller<\/em>How the human body could inform chip design<\/h2>\n\n\n\n
Toward designing chips that use AI without needing the whole internet<\/h2>\n\n\n\n
About Purdue University<\/h2>\n\n\n\n
Papers<\/h2>\n\n\n\n
2024 Design Automation Conference
DOI: 10.48550\/arXiv.2406.18791<\/a><\/p>\n\n\n\n
IEEE Journal of Solid-State Circuits
DOI: 10.1109\/JSSC.2019.2932852<\/a><\/p>\n\n\n\n
2022 IEEE International Solid-State Circuits Conference
DOI: 10.1109\/ISSCC42614.2022.9731793<\/a><\/p>\n\n\n\n
IEEE Journal of Solid-State Circuits
DOI: 10.1109\/JSSC.2022.3159473<\/a><\/p>\n\n\n\n