Technology Description: The technology proposed will deliver a highly scalable, secure, and resilient data acquisition network and utility of panel-level data, creating a rich immutable crypto-secure data set. The proposed use-case focuses on the potential benefit of such a network paired with a cutting-edge AI platform on BTM Solar farm O&M.
Most BTM Solar farms track only current and voltage at a string (1 sensor per 10 or more panels) or string combiner (1 sensor per 100 panels) level. Once enough cells are defective, causing a significant enough breakdown in string level performance, the O&M crew reacts using infrared cameras and/or ad-hoc infield metering to locate the faulty panel before troubleshooting, leading to vastly inflated O&M budgets.
Data granularity, coupled with the platform data structure’s immutability, greatly reduces O&M costs while enabling non-traditional commercial distribution models, which could eliminate the need for upfront capital for panel-level sensor deployment (M&V).
The technology stack has three components:
1. Sensors: Sunified’s proprietary sensor hardware design brings together the following core capabilities: Voltage, Current, Temperature, Bluetooth I/O telemetry, eSIM with enhanced ECC crypto libraries. Sunified’s current 2-chip pre-production sensor module is integrated into a single chip wafer (aka the UNITY™ chip). This chip-scale integration radically reduces cost and enables simple embedding alongside the PV cell-strings assembled into OEM PV manufacturers’ panels. The Sunified sensor (UNITY™ chip) pre-processes raw IO signal into 3 of 4 data fields, where sensor payload is encoded bitwise into our datagram standard. Data filtering algorithms are enabled (if required) before low-level encapsulation and compression of each datagram. When the sun rises, and the panel energy flows, our Bluetooth application sets up a hardware secure virtual tunnel service to our wireless gateway or Edge Node. The security layers “on-chip” are enabled by Sunified’s ECC bulletproof, a hard binding to the module's MAC address, and Unique ID (UID) device serial number as a crypto anchor. The eSIM patented hardware provides a distributed root of trust necessary to support the derived key material required for the secure wrapper around the datagrams. Each sensor dataset is pushed out to the data concentrators at the edge as the data propagates across the mesh.
2. Communication & Data Aggregation Layer: Each panel sensor sends datagrams every 30 seconds to be accumulated and mathematically woven into a Merkle tree data structure and NOSQL time-series DB. The datagrams travel over a resilient self-healing mesh network. Each panel sensor propagates its neighbor’s datagrams as part of its network layer logic. All these datagrams are verified, decoded, and collected to our Edge Node Gateways, a local bank of networked hardened industrial rated Linux Single Board Computers with solid-state drives and redundant data paths. Confidential Computing methods keep data secure: at rest, in transit, and in-use. The 30-sec intervals are scheduled in pseudo-random time pulse so that no single UNITY chip (node) becomes swamped or flooded with data it cannot handle. The industrial Bluetooth mesh provides a secure wireless fabric used for seamless data acquisition. Data is post-processed into a rich data lake held securely and stored across multi-homed RAID storage NAS. The data pipeline provides a rich dataset, ready for post-processing ML and AI analytics. Insights and Analytics as a Service are to be deployed to a Grid Providers RTU & field sensor database/cloud instances. Our preference is to deploy distributed analytics as a service via Blockchain Edge Node, partnering with various Blockchain Network Partners as premium quality Data Oracle.
This means that Sunified’s Solar ORACLE FEED can act as the de facto “Ledger of Record” for Solar Production. This distributed oracle service (deployed for multiple DER sites becomes the MASTER of all cryptographically signed feeds of on-chain sensor data. Ultimately the ORACLE FEED subsumes SCADA & Equipment IO feeds, data APIs, and event streams. This locally distributed oraclized dataset is anti-fragile and can be used for scheduling & balancing critical energy & logic flows when centralized systems are out of reach because of weather or comms/cybersecurity events. We provide the essential datasets required for Fail-Safe islanding of DER assets.
3. AI-based data analytics layer - Quadrical’s AI-ready solar analytics platform is built on a stacked physical and digital twin model. This model allows real-time analytics and alerting, yield prediction, ingestion from rich data streams as provided by the sensor array, error source detection, and triangulation, and features robust handling of gaps or errors in the data feed. The platform has been deployed commercially at a string and string combiner level for large utility-scale solar farms. With individual panel-level monitoring as made economically viable by the Sunified sensors, it will provide greater insight into the occurrence and cause of faults.
State of the Art and Shortcomings of Current Technology: With current technology, there are some major shortcomings in performance monitoring for BTM solar farms, particularly high sensor costs, limited scalability, and low data security. The proposed solution seeks to solve the shortcomings in the following ways:
1. Cost - Compared to traditional Wi-Fi-based sensors, the UNITY chip reduces costs by communicating via WirePas Protocol, thus eliminating the need for an expensive microprocessor to support an IP stack. Compared to RF-based sensors, which have similar costs per sensor, the BoS for the proposed architecture is expected to be between 20-40%lower than current architectures.
2. Scalability - By using Wirepas protocol on top of Bluetooth, UNITY chips mesh networks overcome Zigbee sensors scaling limitation of 256 per hub, to millions of metering sensor nodes per km2
3. Security - The security for each level in the stack created a highly secure system. Each sensor is tamperproof, with a private key, cryptographically generated after the “Proof of Install” activation event stored on the blockchain database. This makes it impossible for bad actors to hack it. The hardened industrial Bluetooth mesh used in the proposed solution has a reduced attack surface compared to the large attack surface over devices with TCP/IP Stacks, Modbus (RS485), or PLC. Finally, each data stream is stored on a blockchain, making its history cryptographically immutable forever and thus much more secure than legacy centralized databases.
Overcoming the Shortcomings, Limitations, and Challenges: The innovative architecture of the proposed solution provides several unique advantages:
1. Granularity - the UNITY chip collects data at the panel level. Currently, most solar farms collect data at the string or combiner box and/or inverter level at 5-minute intervals
2. Data - high resolution, panel level, PV Temp data, along with IV data, provides a forecasting signal for predictive maintenance and PV panel warranty escalation or fault detection.
2. Resilience - wireless mesh networks are self-healing as opposed to wired sensors of closest competitors (two vendors Huawei FusionSolar (string level, poll-only monitoring with comms over power line comms PLC) or Sunsniffer.de (all wired, on panel, Modbus (RS485) sensor, also not a self-healing topology)
4. Pro-Active Reporting - Local data set (held in a secure buffer) on the UNITY chip /Sensor runs a real-time ML signal that can pre-process the raw data signal to escalate on data anomalies, pop an alert or trouble signal should it’s performance be seen as suboptimal. This allows for real-time insights instead of post hoc or a reactive delayed response based on metadata analysis.
5. Anti-fragile - Quadrical’s AI engine will develop the capability to create schemas to fill in any signal or data gaps that might arise due to sensor level attacks or failures
The target performance metrics for the proposed technology are as follows:
Potential Impact of the Proposed Project:
The granularity of sensor data and ML forecast will allow Solar farm operators to quickly and efficiently pinpoint defective panels that may fail in the future, even when a single cell appears to be working. Operators perform preventative maintenance rather than corrective reactive maintenance. Proactive & preventative maintenance optimizes O&M budgets while maximizing operational efficiency. Sunified’s secure and immutable data can generate a host of non-traditional value streams that could transform solar farms’ economics and how they’re financed and operated.
Finally, this project has a real chance of solving the Denial-of-service attack. With Quadrical AI’s platform and Sunified Inc’s panel level data and uber-secure self-healing sensor net, any gaps in data feed caused by a targeted denial of service attack can be compensated for. This project will function as a testbed to perfect the necessary schemas and processes.
Key Technical Risks/Issues: Below are some key technical risks and how we plan to mitigate
1. RF Comms Network Disruptions: Wirepas comms network will face issues like RF pollution and Radio shadows cast by temporary obstructions, leading to signal disruptions. Both of these problems are being addressed by the aid of the Quadrical AI platform which has been specifically designed to support distributed architectures while being robust against signal disruptions, and in the long term by upgrading chips to also include a signal strength data payload.
2. Data Synchronization: Data synchronization is necessary for system-level analytics. While any distributed wireless network with robust retry capability like Wirepas necessarily delivers packets in an unsynchronised (out of order, or delayed) manner, all Sunified sensor packets contain accurate timestamp information allowing reconstruction. Additionally, while waiting for delayed data to arrive Quadrical’s digital twin technology gives us the ability to temporarily fill gaps allowing continued real-time analysis and reporting. Testing the limits at scale is a goal of the investigation.
3. Data Sanity: While the data generated by the Sunified sensors is integrity protected, and the Quadrical platform contains anomaly detection capability to manage anomalous data because of component failure or sensor error. The project aims to understand whether system-level feedback can improve data sanity.
As the system scales to 1,000,000+ sensors there are likely to be several operational edge cases that will present themselves. The proposed tech solution aims to address such cases by having a highly modular and flexible architecture.
Impact of EERE funding: EERE funding will have a significant impact on our project. Establishing a robust testbed implementation with our utility partner will generate key environmental and operational datasets needed to validate, optimize, and refine the single-chip sensor design, mesh network, and AI analytical layer. Reducing the sensor architecture to a single chip implementation is an expensive undertaking that yields a massive cost reduction on a price per sensor basis. EERE funding support would send a very positive signal to private capital to further accelerate that investment cycle. One of the greatest benefits of EERE funding is industry visibility and validation of the dataset we generate and the opportunity to open doors for other companies to explore applications that can be made possible with the immutable dataset: Such applications may include green energy incentive programs from utilities and regulators and higher-value applications such as Proof of Origin, PV Panel Warranty Tracking/Escalation, Predictive Generation with Green Signals, firming dispatch-ability off a DER and the deployment of an interactive ML/AI Grid-Scale Energy Balancing logic and architecture.
Addendum: Business Point of Contact, Geoffrey Cairns, has 30+ years of experience and held senior management positions in Hewlett-Packard and Microsoft, where he was responsible for building and managing complex partnerships and delivering large scale solutions across industry verticals. Technical Point of Contact Lead, Leon Vandenberg, is an accomplished Systems Design & Cryptography Sub System & Network engineer. He has designed, built, and delivered major nation-wide & multinational projects over his 30+ year career. Notable Examples: US Navy Cash for US Treasury, & Navy Supply Office: A global antifragile smartcard e-Money solution. Various DoD wireless & embedded Systems; SIM Authentication Wi-Fi Gateway - edge routing protocol merging British Telecom internet & Vodafone 3G networks via an elegant home gateway, BT Fusion. Key Technical Lead, Michael Kadonoff, brings 15+ years of experience delivering platform products to market. Michael is an industry expert, patent holder, and proven innovator of advanced wireless measurement systems. After starting his career at GE’s Grid IQ division designing grid-scale protection, measurement, and control systems, he then founded, launched, and listed Braingrid.io which delivered targeted sensors & energy and monitoring solutions for residential solar, industrial IoT, and Agribusinesses.
Sunified has assembled a consortium of companies (Quadrical, Avnet, Noether, IBM, and Wirepas) to help deliver the complete solution around the sensors and is actively negotiating with several potential utility partners to provide the testbed for this proposal. The core sensor technology and architecture are protected by 36 issued patents. The company is sufficiently funded to meet the mandated cost-sharing requirements.