Team Members

Retrofit Smart Metering of gas meters in Spain.

The European Union is currently pushing for 80% smart meters by 2020 and is willing to commit up to 45 billion dollars for the effort. One of the biggest areas of interest is in gas meters. Almost every household in Spain has metered gas connection. There is a consensus amongst utility companies that these meters ought to be made "smarter."

The challenge the industry faces:

Employees are sent every 2 months in Madrid to manually read every house’s meter. The cost associated with having an employee do this comes down to approximately $3 per meter per house every year in Madrid (and up to $10/meter/year in other parts of Europe). The challenge of managing a large workforce along with overheads and the annoyance and disturbance this causes to the home-owners (clients of the utility company) create a compelling business pull for IoT solutions.

Most houses are already equipped with analogue meters. For safety purposes, regulation tends to favor gas meters that are purely analogue in nature as there is no risk of explosions. This poses a unique challenge as smart meters require electrically powered sensors.

Therein lies an engineering challenge - how to make a smart meter that does not require electricity and complies with regulation.

There are multiple players in the IoT space for smart meters:

Fully integrated smart gas meter.
Not compliant with Spanish/most EU regulation
Expensive. Currently, analogue meters cost $30 and reading is $3 per meter/year. Traditional smart meters cost above $200 with $100 installing cost.
Present IoT sensors that can read meters
Traditionally deployed on electric meters to read them from a distance. However, they all require power/I2C connection port to read a sensor in the meter and then overlay a communication channel to relay the information remotely.
These cannot work with gas meters as they are purely analogue with no power.

A cost-effective solution is still required.

IoT solution:

We propose a cost-effective IoT platform that enables us to build on top of any existing gas meter in Europe and transmit readings on a daily basis. The IoT platform builds directly on top of research currently being done in Professor Zach Manchester’s lab at Stanford on Sprites (micro-satellites deployed earlier last year from the International Space Station).

The IoT node will consist of a communication layer (LORA), a camera, a microprocessor and power circuitry. The node we aim to build will be under $20 and the camera on board will be able to take images periodically of the analogue output of the gas meter and using a pre-trained computer vision model, we can perform OCR on-chip with little power consumed. The readings can then be relayed using LORA to our base radio station either daily or weekly, as desired by the customer. The device will be powered through mass-produced coin cell batteries. We hope to obtain at least 5 years lifetime without having to replace batteries and all the components are off the shelf PCB compatible components.

Competitive edge of our solution:

Computer vision on-chip. To perform the required functionalities, we need to be able to do Optical Character Recognition on-chip in an energy efficient manner. This will involve pre-train a model that works with the desired gas meters and can run on-chip with as little power consumption as possible,
Powering for an extended period of time: The sensor implemented on top of the gas meter need to continue operating for a five-year lifespan and we can meet this by utilizing coin-cell batteries in series.

Low-cost production: to make the solution economical, we need to reduce the price of the system to $20. Using off-the-shelf products, with optimization and PCB fabrication facilities in China for larger scale printing and assembly, we have been able to reduce the price to $20.

Low power transmission: as per gas utility client’s requirements, we hope to read the meter once a day and transmit it once a week using low power LORA protocols.

Value added:

Besides the cost savings of not having to manually read the gas meters regularly, we also add value by being able to help utility companies comply with current regulation. In addition, for the first time, an accurate dataset of gas consumption on a daily basis with household level granularity will be made available. This dataset can be analyzed for useful applications such as gas leak detection, tracking consumption trends and changes etc. which can be utilized by local governments and agencies for energy planning. Lastly, this can be expanded to other gas/water meters around the world.

Our team:

The core team consists of Professors at Stanford, electrical engineering and computer science graduate students at Stanford along with business partners and developers based in Madrid and India.

Current status:

We are in the final phases of our prototype and will be demonstrating to clients in Spain in August, 2019.



Anand was interning in the summer of 2018 in Madrid at a data science consulting company called The company was approached by a client - Madrileña Red de Gas stating the problem of making their conventional gas meters smarter. As the only engineer with hardware background in the company, Anand was given the project. Anand, while doing research in Professor Manchester's lab at Stanford, was able to connect the technology being developed for satellites and space to the problem being faced by utility providers across Europe. Using the low cost micro-satellites and communications being developed, along with leveraging state of the art low-power on chip machine learning for optical character recognition, the Stanford team has found an economical and technologically viable solution using IoT. The need for an IoT solution that is low cost and can be deployed on any meter with very little customization is very much present - especially in Europe and Asia. The data generated from these meters can be mined for insights that are tremendously useful to local governments and agencies. Overall, such a solution will help improve our energy infrastructure, increase efficiencies and reduce over all costs and over heads.


Voting is closed!