A Living Network of Everyone and Everything

The Berkeley Ubiquitous SwarmLab was formed in 2010 focusing on the transformational opportunities offered by trillions of networked motor-sensory nodes distributed throughout our physical world. To enable the full potential of the sensory swarm, we realized that a degree of virtualization was necessary enabling applications to be developed and deployed on an abstracted model of the hardware platform called the SwarmOS, while still being able to provide real-time guarantees. Many of these concepts are now migrating into the commercial and industrial domain.

Over time, these ideas morphed into something even more daring and futuristic. Advances in every aspect of information technology enable entities such as robots, drones and UAVs to becoming more ubiquitous, and efficient, enabling them to interact more effectively with each other and/or with humans. These networks of everyone and everything create entirely new options in so many areas such as augmented reality, tactile inter-networking, distributed manufacturing and smart cities. To go even one step further, the miniaturization of motor-sensory functions enables the swarm to encompass the human body (“The Human Intranet”) opening the potential of true augmentation, and fundamentally changing how we humans interact with ourselves and with the world around us.

Readings:
- IoT++ - The Living Network of Everyone and Everything (2018)
- The Swarm at the Edge of the Cloud (2014)
- The Swarm at the Edge of the Cloud - A New Face of Wireless (2011)

Advances in microscopic sensing, ULP processing and communications have led to brain-machine interfaces that may be able to observe thousands if not millions of active neurons in vivo. These “nanomorphic” circuits truly push the limit of nanometer scale semiconductor devices., and undoubtedly enhance the prospect for viable long-term brain-machine interfaces, in which sensor nodes directly observe and excite neural activity in the brain, and use this information to restore function for people with severe neural disabilities such as stroke, spinal cord injury, ALS, epilepsy, etc.

Our efforts resulted in the development of the groundbreaking devices such as the WAND and “neural dust”. This has led to the realization of start-up companies such as Cortera Neurotechnologies and iota. Our technologies have also been adopted by high-visibility ventures such as Neuralink.

Readings:
- Brain-Machine Interfaces - From Concept to Reality (2017)
- Wireless Recording in the Peripheral Nervous System with Ultrasonic Neural Dust (2016)
- Brain-Machine Interfaces - The New Frontier to Extreme Miniaturization (2011)

In the late 1990’s, a number of researchers envisioned the tremendous opportunities that would be enable by the availability of large networks of small wireless sensing devices spread throughout our physical environment. Initially called “wireless sensor networks (WSN)”, the possible societal impact became rapidly clear through applications such as smart homes, power distribution, mobility, environmental monitoring, agriculture, and health care. WSNs later became popular under names such as Societal Information systems, IoT, Industry 4.0 and swarms. Getting there required many major breakthroughs to be realized. In the early 2000s, for instance, an untapped opportunity in the realm of wireless data lies in low data-rate (<10 kb/s) low-cost wireless transceivers, assembled into distributed networks of sensor and actuator nodes. While the aggregate system processes large amounts of data, individual nodes participate in a small fraction only (typical data rates <1 kb/s). These ubiquitous networks (PicoNets) require that the individual nodes (PicoNodes) are tiny, easily integratable into the environment, and have negligible cost. The PicoRadio Project tackled the challenges and opportunities in the design of integrated wireless sensor and actuator nodes, to be used in such self-configuring ad-hoc networks. To be viable, the node had to be smaller than a couple of mm/sup 3/, cost <$1, and consume <100 μW, allowing for energy scavenging from the environment. These metrics, which seemed extremely hard in 2000, are now fully realized in industrial settings.

Readings:
- PicoRadios for wireless sensor networks: the next challenge in ultra-low power design (2002)
- PicoCube: A 1cm3 sensor node powered by harvested energy (2008)

In the 1990s, it became obvious that the traditional general-purpose single-CPU paradigm was running into a power wall. Proposed solutions included the introduction of parallelism (as was advocated in the InfoPad project ), and the introduction of specialized processing units (now called accelerators). Pursuing the latter was the intent of the Pleiades processor approach, which combines an on-chip microprocessor with an array of heterogeneous programmable computational units of different granularities (called satellite processors) connected by a reconfigurable interconnect network. The microprocessor supports the control-intensive components of the applications as well as the reconfiguration, while repetitive and regular data-intensive loops (called kernels) are directly mapped on the array of satellites by configuring the satellite parameters and the interconnections between them. Synchronization between the satellite processors is by a data-driven communication protocol in accordance with the data-flow nature of the computations performed in the kernels. A generalized interface wrapper is placed around each satellite processor to comply with the communication protocol. This spatial programming approach resulted in energy efficiency levels at least an order of magnitude better than what can be accomplished in comparable DSP processors and two orders of magnitude over general purpose approaches. More than two decades later, this approach is now has become popular in Systems-on-a-Chip designs (as used for instance in smartphones) as well as data centers.

Readings:
- A 1 V Heterogeneous Reconfigurable Processor IC for Baseband Wireless Applications (2000)
- System-on-a-Chip: A Case for Heterogeneous Architectures (1999)
- Reconfigurable Processing: The Solution to Low-Power Programmable DSP (1997)

The InfoPad project was started at UC Berkeley in 1992 to investigate the issues involved in providing multimedia information access using a portable, wireless terminal. It quickly became clear that a key design constraint was the energy consumption, which could best be addressed through an integrated system approach. The project was therefore organized to address all design levels, including the applications and user interface, backbone network protocols, software for distributed network support, the wireless link, and the pad itself which used a number of low voltage ASIC designs and a processor running embedded code. Tools were developed when not available (particularly in support of low energy design), as well as an interface to mechanical designers who created a custom injection molded case.

As a forerunner of the tablet concept, the Infopad project pioneered many groundbreaking ideas such as: the use of concurrency and parallelism to improve power efficiency, keyboard-less user interfaces, mobile wireless networking, remote servers connected to a “backbone” to provide speech recognition and video services. It ultimately led to the formation of the Berkeley Wireless Research Center.

Readings:

- Infopad and Beyond (1999)
- A low-power, lightweight unit to provide ubiquitous information access application and network support for InfoPad (1996)
- Research Challenges in Wireless Multimedia (1994)

The early 1980’s witnessed the advent of the VLSI revolution advocating a clear separation between the design and manufacturing of digital integrated circuits. The accompanying abstractions made it possible to envision design to become an organized and structured compilation process automatically translating high level descriptions into manufacturable layouts. Our combined efforts in this field over the years resulted in some key contributions that were recognized with awards and resulted in commercially adopted tools. The LAGER layout generation system, developed in Berkeley in the mid-1980’s, allowed for the automatic generation of the layout of multi-processor systems from assembly language descriptions., and was used for over a decade by a range of academic institutions. The subsequent CATHEDRAL systems, conceived a imec in Belgium (which was just newly formed), demonstrated the effectiveness of the approach for the domain of digital signal processing. The tools were adopted by commercial CAD tool companies such as Silvar-Lisco and Co-Ware. Finally, the subsequent HYPER system at Berkeley introduced compiler-style optimization techniques illustrating that automated designs could outperform human designers in terms of performance and energy efficiency. Two of the publications on these efforts were among the stellar group of 42 papers, recognized in the Best of 20 Years of ICCAD in 2003.

Readings:
- Hyper-LP: A System for Power Minimization Using Architectural Transformations (1992)
- CATHEDRAL-II - a computer-aided synthesis system for digital signal processing VLSI systems (1988)
- An integrated automated layout generation system for DSP circuits (1986)

The concept of the switched capacitor circuit was invented in the late 1970s, enabling complete filter systems and analog processing to be realized in MOS integrated circuits. The approach found rapid acceptance in the industrial community, enabling for instance the realization of integrated modems. One challenge however was the lack of simulation tools to help not only model the functional behavior of the circuit, but also estimate the impact of second-order facts such as switch resistance and finite opamp gain. Also of essence in analog circuitry is the capability of modeling noise and non-linear effect. The DIANA-SC program (the PhD project of Jan Rabaey) was the first of its kind to include all these effects in an integrated simulation environment. The program was distributed broadly, and formed the base of the SWAP program suite, commercially offered by an EDA startup company called Silvar-Lisco.

Readings:
- DIANA-SC : a versatile top-down analysis tool for switched capacitor circuits (1983)
- On the simulation of switched capacitor filters and convertors using the DIANA-program (1979)