The Sound of a Smart City

The MONICA project is demonstrating how IoT technologies can manage both sound and security at large, open-air cultural and sporting events – and keep everybody happy.

When Keith Richards sends the first notes of “(I Can’t Get No) Satisfaction” out over the screaming crowd of Rolling Stones fans, you want those ticket holders to feel like they’ve indeed gotten every bit of their money’s worth. But you certainly don’t want it to be at the expense of dissatisfying the neighbours. How can you tell what portion of the sound wafting through city streets and sidewalks is actually coming from the concert, and how much is just the ‘normal’ noise of a pulsing metropolis?

That’s where the MONICA project comes in. A large-scale demonstration of how cities can use the Internet of things (IoT) to meet such challenges, MONICA deploys a cloud-based plat-form to wirelessly connect various IoT-enabled devices. Control systems monitor the data collected and can automatically induce required actions based on the information gathered.

One hoped-for outcome of the project is an acoustic system that can reduce low frequencies outside a concert area, while not interfering with the music in the audience area of venues that use it. It should be able to support designated quiet zones within the concert venue as well. The acoustic system will be automatically adjusted for changes in weather, audiences, music types and other variables.

Twenty-nine partners from around the EU

Brüel & Kjær is one of the many partners involved in the MONICA project. The company’s role is threefold: 1) to create and deliver IoT-enabled sound level meter prototypes for use at the EU pilot sites; 2) to develop the complex algorithms that can accurately estimate the impact of the different sound contributions; and 3) to automatically detect sounds that could indicate a ‘security situation’.

“It’s going to enable not just Brüel & Kjær, but also all the other partners in the project, to collect the data they need – it will be possible to deploy our solution anywhere,” says Brüel & Kjær Research Engineer Karim Haddad, PhD. All data is protected to ensure privacy, and MONICA will comply with the applicable national and EU regulations on data protection, privacy, informed consent and authorization.

LEARN MORE
BRÜEL & KJÆR MICROPHONES

Karim adds: “The sound level meters will contain GPS info, so you can recognize exactly where they are and collect the data you need in the cloud.”

As part of the process of data collection, microphones are placed wherever sound optimization is required and according to the landscape’s unique demands. Anything from the weather, to the location of surrounding buildings, to the sounds themselves affect how sound propagates. “If there’s a long stretch without buildings near the concert area, then the sound waves can travel far distances. But behind a building, there’s not much sound contribution, so you might not hear the concert at all. Temperature, wind and humidity can also affect how sound travels, and some noises carry farther distances than others. It’s quite complex,” says Wookeun Song, PhD, Brüel & Kjær Research Engineer.

Fredagsrock (Friday’s rock) at the Open Air Stage in Tivoli

Is it concert noise or city noise?

LEARN MORE
BRÜEL & KJÆR SOUND LEVEL METERS

Of course, measuring sound levels is only useful in the context of the MONICA project if it’s possible to detect how much of the overall sound is from concert activities and how much is city noise.

That requires Brüel & Kjær to create algorithms, placed in the cloud and linked with IoT-enabled devices that can calculate the different sound contributions.

Some of the algorithms depend on time synchronization between the sound level meters – within virtually a millisecond. Otherwise, it’s impossible for the algorithms to determine how sound actually propagates between different locations. And decisions about what actions need to be taken to ensure optimal sound depend on the availability of accurate data.

“Distinguishing between different sound sources is not an easy task. Furthermore, no one has created this type of algorithm distinguishing concert noise from city noise before,“ says Wookeun.

Brüel & Kjær uses machine learning, as well as other types of algorithms, to solve the basic problem of identifying concert noise vs city noise. During the ‘training’ phase of machine learning, a learning algorithm enables the system to determine how to distinguish concert noise from other types of noise.

Karim Haddad and Wookeun Song work in Brüel & Kjær’s Innovation Team – the team responsible for researching new methods and technologies that solve and simplify our customers’ existing and future sound and vibration challenges

 

A guinea pig in Tivoli Gardens

To create the vast amounts of data necessary for machine learning, Brüel & Kjær has already monitored hundreds of hours of sound at different times of year in Copenhagen, Denmark. This includes different weather conditions, as well as an entire season of Friday Rock concerts at Tivoli Gardens in downtown Copenhagen.

Tivoli, one of the world’s most popular amusement parks, wants to be both a great place for concert patrons and a great neighbour, which makes their interest in contributing to the project a natural fit.

Brüel & Kjær is analysing the data now and is taking even more measurements during the new season of Tivoli Friday Rock concerts, which begins this month. This will enable validation of the algorithm.

Six pilot sites, including Tivoli, are included in the MONICA project. Where the sound level meters will make their formal debut is unknown as of the Waves publication date.

Safe and sound

Security is another aspect of the MONICA project in which sound plays a role. Although video cameras are used at concert venues, they are only as useful as what they can see – which sometimes is a shoulder-to-shoulder ‘wall’ of people.

Sound, however, has no line-of-sight issue, so we can detect acoustic abnormalities instead. Say, for example, there’s a fight going on among some concertgoers, but it’s not visible from a video camera. If there are sound level meters deployed throughout the area, it is possible to determine that there’s trouble near a specific microphone and dispatch security personnel to the location.

By combining video and audio, the picture of what’s actually happening at or near the concert area becomes even more accurate.

Smarter cities, smarter residents

Devices such as smart wristbands, video cameras, loudspeakers, mobile phones and smart glasses will also be part of the portfolio of applications MONICA will be able to offer to enhance city services.

The project has the potential to be used in a wide variety of ways. Based on open standards and architectures, the platform can be reused across multiple applications, with only the application layer specific to the deployment setting.

Look for results of the project in a future issue of Waves.

 

---

About MONICA

The MONICA project stands for Management of Networked IoT Wearables – Very Large Scale Demonstration of Cultural Security Applications. A three-year project co-funded by the European Commission, it involves 29 partners in nine countries. In Denmark, participants include Tivoli Gardens, which represents an amusement park with outdoor concerts; the Technical University of Denmark, which is optimizing and predicting sound transmission; and Brüel & Kjær, responsible for detecting, measuring and analyzing the sound data.

The first demonstrations of the project are expected in Spring 2018, and the six pilot sites are located in Bonn, Copenhagen, Hamburg, Leeds, Lyon and Torino. 

See more: Monica Project Website

This is not the end of the story.
Click here to read about the MONICA project's real-world debut in Torino, Italy.

---

Can it really be that difficult?

One of the many challenges in distinguishing between concert and city noises is distinguishing low-frequency noises from each other. Because low-frequency sounds like a bass drum, bass guitar, car, truck or thunder have a far narrower frequency (Hz) range than high-frequency sounds, they are harder to differentiate – both for the human ear and for a machine.

The Brüel & Kjær solution uses machine learning to detect, first of all, whether the sound being measured actually does come from the concert or not; then, the algorithm determines how MUCH of the total, synchronized sound picture is from the concert.