Image

About


Buildings are evolving into smart organisms through their unmatched concentration of distributed sensing, actuation and intelligence. Indeed, the regulatory decree 2010/31/EU (European Parliament) requires building automation control systems in tertiary buildings by 2025. Still, despite massively deployed sensors of all kind, instead of actual awareness, nowadays at most unconscious processing (C0) is reached.

SUST(AI)N derives theoretical & experimental underpinnings to combine novel distributed intelligence, unpre[1]cedented sensing accuracy, and reconfigurable hardware in a smart building context into a conscious organism that achieves self-awareness through probabilistic reasoning across its connected sustainable devices. SUST(AI)N constitutes the first concentrated effort to explore novel advances in distributed intelligence, reconfigurable hardware, and environmental sensing to establish awareness for smart buildings that reaches global availability of information (C11; through data aggregation across connected reconfigurable hardware), and self-monitoring (C21; via distributed probabilistic intelligence and the sensing of group sentiment). We simplify intelligent building hardware and systems by exploiting electromagnetic signals jointly for backscatter communication, energy harvesting, physical-layer computation offloading, and non-intrusive sensing. Reconfigurable intelligent surfaces are used to support each of these functions.

SUST(AI)N achieves awareness along three high-risk, complementary paths:

1: Reconfigurable intelligent circuits: fit to awareness need by post-installation hardware adaptation
2: Distributed, self-organizing global intelligence: awareness through probabilistic reasoning
3: Unprecedented self-awareness through ubiquitous radio sensing: group-sentiment recognition

It achieves sustainability via demand-tailored adaptive hardware, energy and data-efficient AI,non-intrusive RF-sensing, energy harvesting, multi-party encryption.

SUSTAIN objectives


Buildings are evolving into smart organisms through their unmatched concentration of distributed sensing, actuation and intelligence. Indeed, the regulatory decree 2010/31/EU (European Parliament) requires building automation control systems in tertiary buildings by 2025. Still, despite massively deployed sensors of all kind, instead of actual awareness, nowadays at most unconscious processing (C0) is reached. SUST(AI)N derives theoretical & experimental underpinnings to combine novel distributed intelligence, unprecedented sensing accuracy, and reconfigurable hardware in a smart building context into a conscious organism that achieves self-awareness through probabilistic reasoning across its connected sustainable devices. SUST(AI)N constitutes the first concentrated effort to explore novel advances in distributed intelligence, reconfigurable hardware, and environmental sensing to establish awareness for smart buildings that reaches global availability of information (C11; through data aggregation across connected reconfigurable hardware), and self-monitoring (C21; via distributed probabilistic intelligence and the sensing of group sentiment). We simplify intelligent building hardware and systems by exploiting electromagnetic signalsjointly for backscatter communication, energy harvesting, physical-layer computation offloading, and non-intrusive sensing. Reconfigurable intelligent surfaces are used to support each of these functions. SUST(AI)N achieves awareness along three high-risk, complementary paths: 1: Reconfigurable intelligent circuits: fit to awareness need by post-installation hardware adaptation 2: Distributed, self-organizing global intelligence: awareness through probabilistic reasoning 3: Unprecedented self-awareness through ubiquitous radio sensing: group-sentiment recognition It achieves sustainability via demand-tailored adaptive hardware, energy and data-efficient AI,non-intrusive RF-sensing, energy harvesting, multi-party encryption.

Key Work Packages


WP2 Embedded probabilistic node-level intelligence

WP2 develops a custom intelligence block to be integrated into the sensor node. It will contain a custom-designed accelerator for energy-efficient probabilistic learning, controlled by a RISC-V microcontroller. Success criteria: Achieve target average power over 24 hours during normal smart building operation Exploitable results: Low-power RISC-V processor with node-level intelligence

WP3 Distributed intelligence

WP3 deals with the design and implementation of computationally lightweight AI tools, based on black-box and transparent models (and combinations thereof), reinforcement learning and metaheuristics. KPI: Auto-select the most suitable sensor inputs and deep learning overlay and recover from up to 20% spontaneous node failure or malfunction. Exploitable results: Self-healing distributed smart building intelligence

WP4 Communication and sensing

WP4 investigates integrated RFsensing and communications in a building environment.Four technologies, passive narrow band short range (such as e.g., RFID), passive medium band long range (e.g., LoRa, 802.11ah), wider band active short range (e.g., Wi-Fi), and the supporting reconfigurable intelligence surface will be leveraged to perform integrated wireless communication and human sensing in different scenarios. Human identification & human/object location and tracking will be achieved using RFID-based energy-efficient sensor network. For room-level human sensing and vital sign monitoring, Wi-Fi technology will be applied. LoRa systems will be employed to connect smart objects and to sense human presence. The supporting reconfigurable intelligent surface improves wireless signal coverage and thus realizes RF sensing in non-light-of-sight scenarios. Integrated wireless communication and sensing for different distances and applications With passive sensing technology, fulfill presence detection, in-building human localization and tracking With Long-range technology, fulfill the function of long-distance wireless communication and awareness sensing in large/complex spaces (lobby, underground parking lot) With wide-band and/or high-frequency technology, achieve room-level human activity recognition and vital sign monitoring Success criteria: For the scenarios localization, gesture recognition and sentiment sensing, we will generate simultaneous sensor data from 30+ participants and demonstrate equal or superior sensing accuracy. Exploitable results: Highly accurate environmental RF-sensing for smart-building relevant applications

WP5 Energy harvesting

WP5 deals with the energy harvesting part of the sensor nodes through three different energy sources: light, thermal energy and RF energy. Inside buildings, light mainly comes from artificial lights, making light harvesting challenging due to the low light intensity, compared to outdoors. Thermal energy will be harvested from heaters and other heat/cold sources present in the building.Finally, RF energy will be harvested from the emitters used in WP4, such as RFID, LoRa and WiFi. In all cases, the harvester will include all the stages of an energy harvester: the energy transducer (solar cells, Peltier cells, and antennas), the energy processing circuits, and the energy storage units. Success criteria: energy-autonomous operation of nodes for 7+ days in the Aalto collider building Exploitable results: Energy-autonomous sensor node for smart building contexts

WP6 Security

Data privacy through behavioural biometrics based secure encryption using Garbled Circuit Protocol integration into the sustainable distributed sensing system will be verified in WP6. Unlike many types of physical biometrics, behavioural biometrics can often be gathered with existing hardware, needing only software for analysis. That capacity makes behavioural biometrics simpler and less costly to implement. According to various adversarial cases. In particular, we will consider adversaries with varying capabilities (strong: access to the building automation control system, medium: system knowledge and technical competence, week, worker or inhabitant in the building with no specific knowledge about the system, shallow: external occasional visitor without knowledge about patterns or procedures in the building)and resources(strong: nation-state capabilities, medium: corporation, weak: individual)to thoroughly evaluate the resilience of the system against adversarial attacks. Success criteria: Successful node resilience throughout hackathon conducted in the frame of the project Exploitatable result: Resilient low energy sensor node

WP7 Technology validation in relevant environments

Integration and laboratory-validation of results developed in WP2, WP3, WP4, WP5, WP6 (TRL-4). Success criteria: demonstrate the benefits of the SUST(AI)N technology in 3 benchmark scenarios. Exploitable results: Technology lab validation for C0, C1, C0 consciousness levels

WP8 Communication, Dissemination, Exploitation and
Data management

The objective of WP8 is the integration and laboratory-validation of results developed in WP2, WP3, WP4, WP5, WP6 (TRL-4). The success criteria will be the demontration of the benefits of the SUST(AI)N technology in 3 benchmark scenarios. Exploitable results: Technology lab validation for C0, C1, C0 consciousness levels