Aveiro Living Lab: A communication, sensing, and computing platform for smart-cities

This talk covers the development of a living lab focused on smart cities, presenting research initiatives conducted by the Institute of Telecommunications in Portugal. The speaker discusses various technologies used in smart cities, emphasizing the importance of network infrastructure, such as 5G and local area networks, for enhancing urban mobility, including autonomous vehicles and drone applications. They highlight the challenges facing smart cities, specifically the lack of standardized solutions and collaboration between different urban projects. The presentation details the integration of tools like Kubernetes, Docker, and Grafana in managing data systems and services, as well as the use of computer vision for real-time data analysis. The speaker advocates for an open-source approach and a shift in the perspective on smart city development to promote efficiency and cost reduction.

[music] My name is Dart and uh I'm a research assistant uh in Institute of Telecommunications uh in Portugal. Uh and we have this a living lab uh that I will present to you uh and our research. Uh to start I will present the vis lab. What is the vision? uh what about I think that uh we are doing wrong uh in smart cities. uh after that I

will present the research domains that we work and the technologies uh and also uh the part of the network that we work var communications 5G and also the part of the orchestration the service and use cases and my final remarks uh regarding the team uh the team is the network applications and protocols uh group we work specific more in network applications and protocols um but we have

all of these use cases regarding mobility, drones and swarms, autonomous driving, computer vision. In terms of networks, we work in the 5G, vehicular networks, lo local area networks. um more regarding these topics of software deterministic network, softwaredefined networks, mults edge computing, uh vCloud perception that goes uh connected with autonomous driving and of course everything regarding with real time more regarding the data platform that we have on

the city, the management platforms of of course also these new topics of digital twins uh and also federated learning and starting here with Eclipse Foundation also the softwaredefined vehicle part that we try to bring more this part of determinism of timesensitive networks that we work to the software defined vehicle part. Uh regarding the very living lab one of the points that that I have here is that

I think that we think in the smart cities as software developers and this is the problem for me of smart cities. Um, I don't know how many smart cities do we currently have in Europe, but I think that part of the solutions that we have on one smart city, we cannot push to the others. I think that this is the problem. We don't have this kind of

marketplace that we have some kind of application that connects with the hardware of the other cities. And for me, I think that is the main problem. I put here an example of the Shiong smart city in China that they are building the the city at the same time they are sinking in the the infrastructure of the city in terms of networks and sensing. Uh my point my

point is that one I think that the the smart cities uh should be we should think the smart cities as we sink in the energy uh infrastructure telecommunication infrastructure we need to have a network infrastructure for the city and the sensing part we need to sync on that for the c the city try to some kind standardize that part and after that could build some build some

kind of ecosystem that ecosystem service uh these kind of things. Imagine that you are a company that have some kind of service that analyzes the feed of the video camera. You don't need to deploy another camera. You have already one there. But you cannot do that these days. You only connect like NGSCI data this kind of things. But I think that this will not work and I

think that is not working now. Uh and based on that we have this kind of approach. uh we initial fund this uh network uh but after that we also received some money to invest in the infrastructure uh by urban innovation action also together with the city council of a what we have is we have 44 processing and sensing nodes through 16 kilometers of fiber deployed in the

city um we manage all of these network we also have private 5G networks working only five points but we have there. And also for us, we think that this is the way that we can reduce the capex and opex because if you are a company and if you want to use our sensors, you can deploy your service in our smart city. Okay. Uh and you cannot need

to deploy additional devices for that. You can assess the devices directly and not only buy the software part. Uh and this is what uh we are proposing. Also we also manage all of the infrastructure in a cloud like platform model okay with Kubernetes and other uh tools. Uh and of course as a researcher this is a little bit different but we maintain all of this infra infrastructure

by uh ourselves. Um this is the example of the smart lampost that we have on the top of the smart lamp post we have vlar communications Wi-Fi and also the private 5G. Uh we are also moving the things that we have on the top part of the things regarding the processor to the middle of the smart lamp post because it's easy to make some maintenance. As you

can see to make some maintenance we need to stop the roads uh call to police and also uh contract this kind of companies. Also uh in the post we have millimeter wave and video cameras and also had LAR to to measure the mobility of the city and also to collect some counts. for instance with video cameras [snorts] uh in in terms of research domains and technologies that

we use this is basically more the network part of the city okay uh and I put here some open source projects that uh we use from 5G open 5Gs free 5G camalio camara here we use softwaredefined network to to manage all of the network of the city for instance all of the vlar communications are managed by uh also uh softwaredefined networks we did uh different approach with

open v switch to to to implement V2V. Um also we work with time sensitive networking and the project of open Wi-Fi to do some kind of determinism in Wi-Fi. uh also in the part of the 5G we used uh OCDU now open air interface and these kind of projects with software defined uh radius also for the vehicle communications we used the 82.11p um and for orchestration in

the service we are using kubernetes docker and proxmox I will not specify too much here but as you can understand we have the fiber connected uh with that part of the city cluster that where we have all of the smart posts. This is one of the example of onos. I don't know if everyone of you work with software defined networks but this is the representation of the

city. Uh we I I don't have here but sometimes you will see some vehicles connecting here uh and connects by itself uh the two with v2 v2v uh in terms of orchestration and data platform we also use some of these tools. um we have some kind of uh collection between real time data and nonrealtime data that pass by Kafka. After that we represent everything. We process everything

in some kind of micros service architecture and after that we use this model of fware the Orion to to store everything and also in the mong. This is part of the things that we have. We also monitoring with graphana and prompts. Um also we work with autonomous mobility with v and rovers. Okay, this one this is one example of the pix kit that we use to work

with autoware. Uh and this is the architecture that we use. Uh here we have the onboard units that connect with one of the project that we extend that is called Vanetsa for ETS G5 message. Um we implement all of these message and we also are part of the autoware foundation contributing with the V2X uh stack. Uh in that project as you can see we also work with

the pixock uh in the hovers and in the drones also the use case that we usually do in the in the drones is to extend the network like the 5G uh network try to extend and to to give more cover but but al but also in the part of perception in terms of service and use use cases. This is an example of uh one of the French

companies I think Navia uh that has this kind of autonomous bus. Uh the autonomous bus comm uh communicate with it G5 and we create some kind of virtual smartphone uh that when we with the camera detect the cyclist we send a message to the the Navia bus and Navia bus will stop. The this is one of the examples uh with integration of our platform. uh at the

right uh is only the vision that you can see with the autonomous vehicle. Okay. And based on that we are losing also that autonomous vehicle to work in more in the software defined vehicle part. Uh more things >> the bicycle or also >> we also have works in the bicycle part. Okay. This is the the platform that we implement. This is more for because we don't have

any product with that and we are from research perspective. But this is the city. As you can see, you have the yellow ones is everything that you are detecting with video cameras. Okay? For instance, if you have a Volkswagen based automotive and if you enter in the city, we will see your car your velocity because the the Volkswagens communicate with ETSC5. I don't know uh what is

the the registration of the car. I don't know all of that specifics. But I will see the velocity and I will use the velocity of the car to to to mobility perspective. For instance, that road with that car. I I will uh store that information. Uh these are the smart lamp posts. We can do different use cases. But this is the vision of real time that we

can uh see uh with cameras. We can do a lot of things. Uh this is an example what we can do with the uh computer vision. The smart parking at the top here. Some kind of metrics regarding the smart crosswalk. How how many people is waiting here? what what is the time how much time people spend here and the proximity of the cars there uh to the

people that crosswalk is very problematic is v near the university and a lot of people is always passing there and also can use for for tourism. Um in terms of man monitoring platform we collect a lot of data and sometimes we have a lot of problems because of that because we generate a lot of data. Um I put here we use graphana and prompts to do that

regarding we can see the message that we are uh using uh transferring in the MQTT and in the brokers also in the Kafka sometimes the Kafka imagine that some student do some kind of crazy thing and like put a lot of message in Kafka we can understand that sometimes this happen um more things uh this is the Orion perspective and I I also have here the perspective.

Uh we also store a lot of documents in in is our uh database that we are using. Um as as you can see we have a cluster of Fireware Orion uh service and uh more things here and you can see we are measuring uh all of the performance. This is some examples uh from some verticals that we have on the city. This is the smart parking part.

That one is the the mobility perspective. As you can see, we have like uh all of the the roads we can we could put the segments of the roads and associate uh some of the metrics there. Also, this is one of the simple what we can do with the smart crosswalks. This is a normal day. All of the people at 9 go to the school and that

one is a university night event. As you can see at six hours all of the people is go is are going to home. Um regarding the integration with AI we also in the last year integrate this agent to we can make some questions to the the database and the agent will answer to the to that. Um and that's it. Uh my main points here is that we

need to think uh in smart cities in a different perspective not only in the software software perspective. Uh guys I think that we need to like to create this this standard like like we creating the smartphones. We can have different uh brands of hardware. Okay. But all of the applications are the same. I think that we need to do that to reduce this capex and opex. it

not for me it not makes sense to have two cameras doing the same in the same place collecting the same data okay uh and that's it I think that's I put this me message here that I think that we also can build smart cities with open source because part of software are open source and that's it [applause] thank you thank you so much I don't know if

we have question. Uh as usual, you will be available around so people can reach you. Uh there is one race and okay so you mentioned that it works for Swagen but we are hearing about V2X and V2V and other stuff for like past 15 years. I never seen real use of that. So what do you think? what is the future for you know all these nob whatever

uh >> I think that there are >> yes sorry >> I think that there are two perspectives right um you could have that perspective that Tesla use okay you have the sensors in the car you sense everything you don't need to connect with other vehicles and other devices I think that the the V2X part will complement that uh I think that the main problem is that uh

we in Europe Europe defend one standard the guys in America defend another standard and I think that this is the the main part and after that you have in the 5G some kind of specification for V2X and you have this specification for uh Wi-Fi okay this one from Wi-Fi is is easy to to work for us as research the the the other one from 5G is more

difficult to work. But for instance, Volkswagen implemented Wi-Fi 1. Okay, I think that in the future we will have that. I don't know which level we'll have, but I think that we will have that. >> Okay. Okay. Thank you. >> [music]