Basic information

Title: Be Memories interactive tourist guide in Bristol
Acronym: Be Memories in Bristol
Company: HOP Ubiquitous (HOPU)
Type: SME

Where: Bristol (M-Shed)
When (time-plan): November 2018 to June 2019

What is Be Memories

Be Memories is an interactive tourist guide, co-created by the citizens. This guide uses IoT devices called Smart Spots, which allows transforming the cultural Point of Interest (POI) into a Smart Point of Interaction denominated as Smart POI. These smart areas create a new communication channel between the cultural offer and the citizens/visitors.

The Smart POIs have an area between 1 to 80 meters of range where the user can interact with their Smartphone through Wi-Fi. The device creates an open Wi-Fi network using the city internet network or a SIM card. When the user connects their Smartphone to the Wi-Fi network indicated, a web-app with content about the POI is open automatically.

Be Memories Progressive Web-App

The integration of the Be Memories tool in the ICT Flame platform described in the deliverable 3.4: FLAME Platform Architecture and Infrastructure Specification is based on testing different platforms to provide the needed architecture for the tool, avoiding the use of private platforms. The experiment changed the use of third party servers with the ICT Flame architecture, using the ICT Flame cloud platform, together with the Edge computing and 5G (virtualization capabilities) of the ICT Flame infrastructure. The results of this adaptation improve the readiness for GDPR, building a tool more scalable, improving the quality of experience.

Scheme scenario

Scope/objective of the trial: What are you learn? Technically, and regarding experience?

The data stored have been collected by the event register created for trial. The main outcome is to analyze user experience in both territories and compare the video playing in both infrastructures according to the KPIs:

  • Reduce latency for content provisioning of device communications.
  • Decrease deployment time
  • Expand the multi-source retrieval of the solution.
  • Build a solid connectivity interference-free and an increase in bandwidth.
  • Increase the content security, privacy control and geolocated content provisioning
  • Test the Be Memories acceptance in new environments

The ICT Flame provide HOPU a high level knowledge due to the participation with the different partners of the project, developing the different test and trials during the Open Call period as the use of different technologies in the ICT Flame platform, different infrastructure equipment and components of edge computing and IoT technology:

  • LXD images and containers.
  • Get to know and become familiar with TICK Stack.
  • Discover TOSCA definition.
  • Increase knowledge about SSH and reverse tunnels.
  • Increase capacities to involve people in experiments and tests.
  • Know about the existence of Ruckus outdoor access points.
  • Streaming video formats such as DASH and HLS.

Summary of methodology

The technical process to deploy these virtual Smart POIs in the Flame network consisted in the deployment of a Service Function (SF) with the video storage, two SF with web server to offer the frontend with different content on each tower, and a SF with a backend, which manages all comments and assessment in one of the available cluster. Both towers had a Service Function Endpoint (SFE) which request the videos when a user accesses to play video, installing these contents in the SFE for the following accesses. The videos are downloaded from the storage SFE to be nearer to users If it detects that the videos do not be in local. The three SF were integrated with Telegraf, thanks to this tool, web browsers send all data generated during video playback to CLMC, values such as the following:

  • Time stamp of using
  • Features of device
  • Playback percentage and seconds
  • Video playback interruptions
  • Number of visits to the Web-App

Furthermore of register values in CLMC, the surveys provided the impact and opinion of Be Memories between the Bristol public. This experiment has been compared the results obtained during the trial in Bristol, using the ICT Flame Infrastructure with the deployment of Be Memories in Ceutí. based on SBC devices (Raspberry Pi) which are used to generate the access points with 4G internet access.

Results for completed trials

As the basis of the comparison, it is important to understand the differences of both trials, the Ceutí use case has not a platform, the Smart POIs  are independent Edge Nodes, positioned on different places and access to backend over 4G comercial connection. On the other hand, Bristol deployment uses internal logic of FLAME platform. It is escalable by deployment mechanisms. This infrastructure is more powerful than the Ceuti deployment.

Based on the objectives defined at the beginning of the project, the data obtained show the following results:

  • Reduce latency for content provisioning of device communications: The average of initial load time was lower in the Flame infrastructure than in Ceuti, in Ceutí was 446 ms (≈1/2 s) and in FLAME 435 ms (≈1/2 s). These values have been measured from the loading of the application to the start of the video.
  • Decrease deployment time: The orchestrator deploys the LXD image in less than 10 minutes for several distributed clusters. The Ceuti deployment is made individually for each device and it takes more than 10 minutes.
  • Expand the multi-source retrieval of the solution: With high video quality, the both use cases have maintained the necessary bandwidth for simultaneous video playback with the possibility of 9 devices. According to the quality of video playback in both infrastructure, the Flame use case provide better quality videos than Ceutí. The main reason of that was the infrastructure capacity.
  • Build a solid connectivity interference-free and an increase in bandwidth: During Bristol trial, only 2 of 50 samples have obtained interruptions. The data collected ensure that interruption average time is really low, in FLAME 372 ms (≈1/3 s) and in Ceutí the time is 1285 ms (≈1 s) (image below).
Comparison of the interruption average time between Ceutí and Bristol on video playback.
  • Increase the content security, privacy control and geolocated content provisioning: The number of hops was reduced due to FLAME’s current infrastructure in MShed, customizing content for each position. In addition, the data are contained in platform and it is not owned by private companies such as Google with its private storage or Youtube.
  • Test the Be Memories acceptance in new environments: According to the results of the survey in both trials: In Bristol the 50% of them valued with 5 stars the tool and 34% with 4 stars. In Ceuti, the 56 % of the sample value the tool with 5 stars and it is only one assessment with less than 4 stars. The question about if the sample like this type of channel to see cultural content during their visits the 99% answer “yes” in Bristol and in Ceuti the 100%.
  • The quality of the video 66% of Bristol sample valued with 5 stars and 30% with 4 stars.

About the interruptions of the video in Bristol, the 70% of the users answer the video has been played fluently and the 26% had suffered one interruption. Only the 4% had more than one interruption (image below).

Be Memories interruptions in Bristol

To know more about Be Memories experiment, read also their blog post.