Skip to content

nawaf-m/flow

Repository files navigation

Flow

alt mina_stampede

Inspiration

On September 2015, more than 2,400 pilgrims died during the Mina stampede tragedy. The media labeled it as the deadliest Hajj disaster in history. We want to build a solution that prevents such tragedy from happening again. We want to save lives.

Let the Human River Flow.

Solution - what it does

Flow is a solution designed to equip the Ministry of Hajj and official Hajj organizers with the capability of real-time monitoring of pilgrims’ traffic, the ability to predict and prevent dangerous congestions, and seamless Tafweej scheduling (current mechanism for organizing the traffic of pilgrims) updates.

How it works

alt system_model

Flow is composed of three main elements:

  1. FlowDashboard: a centralized intelligence dashboard. It is a web-based dashboard used by the Hajj organizing entities for monitoring and controlling pilgrims’ traffic.
  2. FlowNetwork: designed as a mesh network and as an independent reliable WiFi network to provide connectivity between the solution’s components.
  3. FlowPoint: a data collection tool (wearable or mobile app) used by the Hajj campaign guides.

How we built it

What Is Done So Far

FlowDashboard: The web-based dashboard that provides centralized intelligence to Hajj organizers has been prototyped. The React JS library has been used for developing the user interface. A mock user interface to simulate the usage of the wearable device or smartphone by the pilgrimage guide has been created to be used during the demo.

alt FlowDashboard

Technology Required for Remaining Components

FlowNetwork: Since the pilgrimage rites are date-specific, all pilgrims will have to be in the same location at the same time. Therefore, to avoid relying on a congested service provider public network, for time-sensitive updates, the Flow Solution rides on a private multi-hop Wifi mesh network (reserved frequency) to relay tracked GPS points and campaign leader information to Flow servers. Ruggedized outdoor access points (AP) will be installed from Makkah to Arafat, approx. 1 KM apart, to facilitate this solution. In the four areas (Makkah, Mina, Arafa, Mozdalifah), we double the number of access points to scale. Specific access points at each zone will be chosen strategically as portals to the Internet. In total, that includes 90 APs. Since the transmitted GPS data is minimal (~ 60 mb per FlowPoint during the 5-day Hajj process), and the anticipated number of FlowPoint users is approx. 10,000 users, these APs will be able to support the load. WPA2 encryption will be used to secure the FlowNetwork infrastructure.

FlowPoint: a simple wearable device with GPS tracking and WiFi chips that will be sending its location to the FlowDashboard frequently, and has a small screen for the Hajj guide to receive instructions on how to proceed. (for the purpose of the demo we will use mobile apps).

Challenges we ran into

Communication

Challenge: There is an immense pressure on mobile networks and the Internet at the 4 Hajj locations, as well as during traveling in between them. Data feeds in real-time, which is the essence of Flow, require a reliable strong communications network.

Solution: Flow communication will ride on a Multi-Hop Wireless Mesh Network, operating on a reserved frequency, resulting in a seamless communication channel. This mesh network will work as the Flow backbone, with ruggedized hardware to withstand extreme environmental conditions and a few strategically chosen portals to connect to the Internet.

Adoption Resistance

Challenge: We knew we could not enforce wearables on all pilgrims, but we wanted to find a way to track pilgrims’ locations, without being intrusive.

Solution: Pilgrims are organized and led in campaigns. Every group of pilgrims is accompanied by a campaign leader, who accompanies them throughout their journey. We chose campaign leaders as a data point, to represent their respective campaign of pilgrims. Since campaign leaders report to the Ministry of Hajj, which the solution Flow is built for, it will be mandated on them to sport Flow wearables. Consequently, campaign leaders are the point of authority on-site, who decide when a campaign starts traveling towards a location to perform a rite.

Accessibility and Ease of Use

Challenge: Campaign leaders are fully pre-occupied throughout the 5-day pilgrimage. Grabbing their attention for time-sensitive alerts can be difficult. Additionally, the less intrusive the tracking and alerting solution is, the better.

Solution: Flow endpoints will be practical wearables that contain GPS trackers and wifi chipsets. Wearables, named FlowPoints, are worn on the wrist to free up hands for multitasking purposes. The Flow team decided to opt out of using a smartphone app as their platform, but are utilizing it in the meantime to showcase the solution at the Hajj Hackathon, due to lack of hardware at the venue.

Accomplishments that we're proud of

  • Developing the idea of a reliable Wifi Mesh Network as the backbone for our solution.
  • Achieving a reliable level of abstraction by consolidating pilgrim campaigns to their group leaders, which reduces the number of users we need to rely on to track the whole pilgrimage and have a working solution.

What we learned

  • You cannot mandate a solution on a mass of people and expect it to work as anticipated. It’s better to create a solution that has an automatic opt-in (a solution based on fingerprints; everyone has fingerprints) than to have a solution where we depend on people to opt-in
  • Some Hajj challenges are related. Building a solution to one challenge in the Hajj experience will be consequence touch upon solving a few other challenges.
  • We were exposed to a more detailed narration of the Hajj experience, from many perspectives (officials, volunteers on-site, organizers, pilgrims), which has helped us have a deeper understanding of the problems at hand.

What's next for K-002-Flow

  • Utilize AI technologies for predictive analysis:
    • Use of AI and search algorithms to predict collisions given the location of guides
    • Use of machine learning - as a time-series prediction problem - to predict the number of pilgrims per hour in a certain location
    • Use of deep learning for image analysis for multiple predictions:
      • Congestion areas
      • Outliers
    • Use of machine learning - as a classification problem - for automating action-taking when alarming scenarios are detected
  • Analyze the flow of people to determine ways to ease the congestion
  • Apply the solution on other crowd control cases, e.g. Tawaf al Wida and beyond Hajj

Judges, we need your vote to make this a reality. We need your vote to save people’s lives.

About

Hajj-Hackathon

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •