Akraino 5G MEC Hackathon: Winners, Insights and More
A few months ago, more than 76 developers gathered to participate in the Akraino 5G MEC Hackathon, sponsored by LF Edge, China Mobile Technology, ETSI and ETSI MEC. Attendees participated in-person at the Qualcomm campus in San Diego, a breakout venue in Columbia University and virtually via Zoom conference.
Different blueprints and solutions will be worked through in real time, including: Micro-MEC, AR/VR, 5G MEC edge stack accelerated with SmartNICs, and more. For the hackathon competition, we also featured a top prize of $1000 and $500 for the runner-up teams. Please read on below for feedback from participants and details for the winning teams – Team Planet (1st), Team BlueHat (2nd), Team Blaz3r (3rd) and Team grind (3rd-tie).
“It was a day trip to San Diego to attend the Akraino 5G MEC Hackathon. I was excited when I entered the conference room in Qualcomm campus. Honestly, I never expected so many people. There were people from all over the world gathered in the room, which made it look and feel crowded in a good way with lots of enthusiasm radiating from everyone. What made me more surprised is that I saw a lot of college students. As an open-source project and cutting -edge (oh yes, edge!) technology, we really need the fresh blood! These students really have a lot of good ideas of how to prosper our edge computing platform, surveillance, IoT automotive, they bring creative and inspiring scenarios to the popular applications cases for MEC. Also, the hackathon is a good platform for people in the industry to change ideas on edge, I know a lot of people and companies got the chance to collaborate through this event. As a sponsor, CMTI (China Mobile Technology USA) also get a lot of good ideas and partners. I have to say it was a wonderful day!” – Su Gu, CMTI 5G ICSV
Some highlights on the winning teams are below.
1st Prize: Team Planet
Participating in Akraino 5G MEC Hackathon is an extraordinary experience for us.
After brainstorming at the beginning, we decided on the direction of our exploration: adopting the edge computing paradigm to mitigate the privacy issue of surveillance in the smart city. The foundation of smart city applications is the enormous amount of data collected from physical space. However, pervasive sensing also raises privacy concerns as the collected data may be highly sensitive. Even worse, the massive adoption of cloud computing in smart city applications makes sensitive data generally processed by the untrusted service provider on untrusted infrastructure. We argue that edge computing is capable of mitigating existing privacy issues as it could provide a different trust model to the smart city applications.
To instantiate our thought, we designed the driven scenario as a privacy-preserving video surveillance application in a meeting room. Video surveillance is commonly used in mission-critical spaces for security and safety purposes, like theft protection, environmental safety monitoring, and emergency response. However, on the other hand, video footage of public physical space is also highly sensitive. Residents are usually reluctant to let the video footage be reviewed or stored unless there is a real emergency or anomaly. So we designed our application, which first sends captured video to trust edge infrastructure to detect if there is an abnormal situation. Only the photos reflect the anomaly will be shared with the space manager on the cloud and be stored. In our proof-of-concept prototype, we define the abnormal condition as higher-than-expected occupancy in a room.
To implement the prototype, we leveraged the edge computing platform provided by MobiledgeX. It gives us a convenient way to deploy a containerized application to an edge infrastructure nearby. We deployed an open-source face recognition at the edge. We then implemented a Python program on a laptop to make it function as a video surveillance camera by using its webcam. The laptop keeps capturing the pictures, sending them to the edge, and calling the face detection algorithm to count the people in the room. If the occupancy is higher than a certain threshold, the image is sent to the cloud and stored. Otherwise, the program drops the image to protect residents’ privacy.
During the final review and judging phase, judges and audiences gave a lot of helpful comments and feedback. We together discussed topics like the scenarios that edge infrastructure is trusted, the capability of enhancing this application using trusted hardware, and how to extend this use case to other situations like controlling the activation of voice assistants.
In this hackathon, we not only learned more knowledge about edge computing but also got hands-on experience on real-world edge computing platforms and opportunities to build connections to the open-source community. Thank you, Akraino, for hosting this fantastic hackathon event. Thank you, Vikram and Bruce from MobiledgeX, for providing the edge computing platform and all the kind supports. Finally, a big thank you to everyone in this hackathon, for sharing your brilliant ideas and insights. We hope there will be more events like Akraino 5G MEC Hackathon that provides students opportunities to learn more about critical and cutting-edge technologies.
2nd Prize: Team BlueHat
When four of us read through the hackathon prompt, we thought to ourselves: how could we leverage a smart city’s sensor network to produce real impact? Our experience living in the city drew our attention to city traffic: very often we see an emergency vehicle, such as a fire truck and an ambulance, getting stuck at a red light behind a long line of cars. If the line was short, cars in front would notice the emergency vehicle behind and would actively run the red light to let the emergency vehicles pass the intersection. However, running a red light is inherently dangerous. In addition, when the line of cars becomes long under heavy traffic, the cars in front would often not be able to notice the emergency vehicles stuck way back. Therefore, we realized that an infrastructure-level solution is needed and developed Smart-city Emergency Express (S.E.E.), a traffic control system for smart cities. Using S.E.E., traffic lights can actively detect emergency on the streets and if those vehicles are found, they automatically switch to green lights for them to pass. The overall hackathon experience was exciting and fun. We have posted our code on Github. Thank you Akraino for the recognition of our work and hosting this meaningful hackathon! – Team BlueHat
3rd Prize (tie): Team TrailBlaz3r
It’s obvious that starting from years ago, the growth personal vehicles has exploded exponentially. As we progress into the future, the problem will only get worse because there are always more new cars than scraped cars. With this problem comes the challenges of parking, especially in the more densely populated area. For example, in major european, asian and north american cities, cars are all over the side of the streets and is extremely difficult to find parking. With the uprising of 5G as well as this device, we envision a situation where all the parking spots are being recorded and regulated such that you would always know, in which area of vicinity there is parking. It is applicable to mall’s parking structure. As we know, parking space indicators are individually placed for each parking spot. Therefore, it’s inefficient because of the amount of sensoring device required. With 5g technology, real-time parking availabilities are viable through prediction model and they could be sent to modile devices so that customers could have first hand information about uhe availability of parking spot. Additionally, with the upcoming of autonumous car, it would be also great for guiding cars without human onboard, to certain parking area in case there is no parking near the driver’s leaving point. These traffic/parking availability information could also be sold to data broker for specified needs.
3rd Prize (tie): Team Grind
On seeing a Facebook post about a hackathon in Qualcomm, San Diego, I was naturally inclined to attend it considering the proximity and value addition it had to offer. After treating ourselves to cookies and coffee, we began brainstorming approaches to solve a problem related to smart cities. After going through the slides and evaluating our options, we finally decided to develop an air quality analytics system. We developed a comparative model between Los Angeles and San Diego to explore possibility of location-specific alert generation system for air quality standards. We learnt the importance of edge-computing and familiarized ourselves with the latest technologies in different companies at the hackathon – Arm, China Mobile etc. I connected with wonderful people like Tina Tsou and Robert Wolff, whom I had the pleasure of meeting again at the Arm IoT Dev Summit in Mountain View, California on Dec 2-3rd, 2019. I hope to stay in touch with the community and contribute towards furthering technology on the edge. ~Nitish Nagesh, 2nd year CSE master student, UC San Diego