• Published on 11 Dec 2015
• External News

Bootstrapping Cyber-Physical Systems via Micro Plug-and-Play: Auto Identification, Discovery & Configuration of peripherals at Extreme Low Power

Creating, deploying and configuring applications for Cyber-Physical Systems (CPS) still remains complex. Micro Plug-and-Play (μPnP) tackles this complexity problem by offering a near-zero configuration approach that empowers non-expert end-users to quickly take their CPS infrastructure online (www.micropnp.com, @micropnp).

Industrial demand for a pluggable CPS at near-zero configuration and energy cost

Through long-term industry collaborations in the domain of transport & logistics, building management and industrial automation, researchers at iMinds-DistriNet/KU Leuven (distrinet.cs.kuleuven.be) learned that CPS technology providers and system integrators lack software management services and tools to integrate, configure and monitor a distributed CPS.

For example, the successful exploitation of a cold chain monitoring solution in trucks and trailers requires the CPS infrastructure to be seamlessly integrated and configured (e.g. every time a truck picks up a trailer), and that fleet management applications dynamically discover the available peripherals (e.g. different sensors, from different vendors, with different capabilities).

Integrating CPS peripherals is time-consuming; the developer must physically integrate the peripheral, manually configure all settings, write specific driver software, and deploy and manage the drivers installed on each CPS device. Mainstream approaches (e.g. USB for regular computing devices) are too heavyweight to apply on battery-powered and resource-constrained sensors or actuators.

Contemporary USB peripherals carry a type identifier that associates the device with a specific driver, which is automatically discovered, deployed and installed. The question remains: how can this be achieved on battery powered and resource-constrained CPS devices?

μPnP solution in a nutshell

μPnP splits the sensing/actuating capabilities from the computing and communication platform; it automatically identifies all connected embedded peripherals on a CPS device and installs their corresponding driver software over the air. μPnP offers a coherent hardware & software toolkit that provides:

1. Low-cost, low-power peripheral identification. The identification approach is ultra low power, low cost, and does not require specialised peripheral hardware. μPnP automatically identifies connected embedded peripherals and installs their corresponding driver software. It utilises passive electrical characteristics to encode a unique 32-bit type ID on each sensing or actuation peripheral. Identifying a peripheral with µPnP consumes 1 million times less power than USB. Any existing IoT peripheral can immediately be repackaged as a μPnP-peripheral by using the standard physical connector (ADC, GPIO, I2C, SPI, or UART) and adding four resistors.
2. Remote peripheral discovery, deployment & usage via standard IPv6 Multicast. μPnP connects IPv6 multicast addresses directly to physical phenomena to enable low-overhead peripheral discovery and remote access. Following the integration of a peripheral, μPnP auto-generates a corresponding unicast prefix-based IPv6 multicast address and joins the matching multicast group (e.g. for temperature, light or humidity sensors). Peripheral services can be remotely accessed using CoAP over both unicast and multicast, allowing for efficient and easy integration with third-party technologies. A domain-specific configuration language was created to achieve platform-independent and extremely compact driver modules (i.e. a few bytes); this minimises the energy overhead when deploying drivers over-the-air.
3. Global web access to configure all nodes remotely. Global access to each μPnP network is provided through a JavaScript library that establishes a NAT-proof connection between clients and the μPnP gateway. This library defines the <pnp> HTML tag, which allows sensors to be embedded in a web page with a single line of HTML code. The library also defines PnP-Script, a rule-based language for non-experts. It can be written in a web browser, and executes on the gateway. This architecture achieves sub-second latency for actuation and control in tests with over 100 simultaneous web clients.
4. Ultra low power mesh networking. The μPnP software stack is fully IETF standards-based (IEEE802.15.4e, IPv6 (6LoWPAN), CoAP, JSON), and therefore integrates easily with third-party solutions. In combination with Linear Technology’s SmartMesh IP, for example, μPnP achieves 10 years of battery lifetime and 99.999% end-to-end reliability.

Key differentiators

•  Ultra-low power (6 orders of magnitude lower than USB)
• Easy driver language (-50% Code, -90% RAM)
• IPv6 Multicast discovery (in 2KB of ROM)
• Extremely low cost (< 1¢ per peripheral)
• Tiny memory footprint (<10% of AVR/ATMega128)

μPnP in action

μPnP was selected as semi-finalist for the IPSO Challenge 2015, which evaluates IoT prototypes on their innovation, breadth of applicability, marketability and ease of use (http://challenge.ipso-alliance.org/challenge2015/meet-the-2015-semi-finalists/micropnp). The ten semi-finalists will present their prototype for an industrial jury at the Designers of Things conference in Silicon Valley, California (2-3 December, 2015).

Beyond the IPSO Challenge, iMinds-DistriNet is taking μPnP out of the lab and into industrial reality. The first step was to equip the Computer Science Department’s server room with μPnP enabled devices. Currently, iMinds-DistriNet is preparing CPS prototypes in transport & logistics and e-healthcare in close collaboration with industry. Clearly, more opportunities are quickly emerging, for instance in industrial automation (monitoring machinery and robots on the factory floor).

iMinds-DistriNet track record in industry collaboration

iMinds-DistriNet is an international research group with extensive expertise in secure and distributed software, including middleware. Embedded in the department of Computer Science at KU Leuven, iMinds-DistriNet has a headcount of over 75 researchers, including 10 professors and 20 senior researchers. iMinds-DistriNet is part of the iMinds Security Department, a de facto “one-stop-shop for ICT security research”. More information: https://distrinet.cs.kuleuven.be.

iMinds-DistriNet has ample expertise in initiating, executing and delivering application-driven research, often in close collaboration with industry partners. Currently, iMinds-DistriNet is actively involved in about 35 regional and European projects, often in close collaboration with industry. The know-how of iMinds-DistriNet was at the basis of Ubizen, a spin-off company specialised in secure e-business and related security services (now part of Verizon).  A second iMinds-DistriNet spin-off company, Qmedit, focuses on the development of medical workflow software (now part of Agfa).

Contact Information:

For more information on our R&D activities (e.g. on IoT security, software configuration management, resource management, and IoT-Cloud integration), contact Dr. Sam Michiels, Industrial Research Manager (sam.michiels@cs.kuleuven.be) or Prof. Danny Hughes (danny.hughes@cs.kuleuven.be).