[GT] Reconfigurable Heterogeneous Integration Using Stackable Chips with Embedded Artificial Intelligence
Imagine a future, where cellphones, smartwatches, and other wearable devices don¡¯t have to be shelved or discarded for a newer model. Instead, they could be upgraded with the latest sensors and processors that would snap onto a device¡¯s internal chip. Such reconfigurable ¡°chip-ware¡± could keep devices up to date while reducing our electronic waste.
MIT engineers have taken a step toward that modular vision with a LEGO-like design for a stackable, reconfigurable artificial intelligence chip, recently documented in Nature Electronics. The design comprises alternating layers of sensing and processing elements, along with light-emitting diodes that allow for the chip¡¯s layers to communicate optically. Previous modular chip designs have employed conventional wiring to relay signals between layers. Such intricate connections are difficult if not impossible to sever and rewire, meaning such stackable designs are not reconfigurable. With the new technology layers can be swapped out or stacked on, to add new sensors or updated processors.
With this design you can add as many computing layers and sensors as you want, such as for light, pressure, and even smell. The researchers refer to this as ¡°a LEGO-like reconfigurable AI chip¡± because it has unlimited expandability depending on the combination of layers.
The researchers are eager to apply the design to so-called edge-computing devices including self-sufficient sensors and other electronics that work independently from any central or distributed resources such as supercomputers or cloud-based computing.
According to the MIT team, as we enter the era of the internet of things based on sensor networks, demand for multifunctioning edge-computing devices will expand dramatically. And in the future, this proposed hardware architecture will provide high versatility for edge-computing.
The team fabricated a single chip, with a computing core measuring about 4 square millimeters. The chip was stacked with three image recognition ¡°blocks,¡± each comprising an image sensor, optical communication layer, and artificial synapse array for classifying one of three letters, M, I, or T.
The simple prototypes demonstrated stackability, replaceability, and the ability to insert new functionality into the chip.
Going forward, the researchers plan to add more sensing and processing capabilities to the chip, and they expect the applications of this architecture to be boundless.
For instance, you could add layers to a cellphone¡¯s camera so it could recognize more complex images or make them into healthcare monitors that can be embedded in wearable electronic skin.
Another application involves modular chips, built into electronics, which consumers could choose to build up with the latest sensor and processor ¡°bricks.¡±
The researchers say it is possible to make a general chip platform, and each layer could be sold separately like a video game. Manufacturers could make different types of neural networks, for image recognition, voice recognition and applications; then let the customers choose what they want, adding to an existing chip like a LEGO set.