Using Technology Solutions for Effective Social Distancing

The global Pandemic of COVID-19 has brought our nation to a staggering halt. The early lockdown proved to be effective in helping us control the situation in the country. However, the spread of the disease has not been curtailed completely yet. Instances of people gatherings at stores and local markets have been a limiting constraint to implementing social distancing. With the government planning to relax restrictions, social distancing will be the only viable containment measure to prevent a resurgence in infections. The cases in our country have been on the rise and the measures taken by the government and the people are going to determine the trajectory of the curve from here on. 

What is the Issue?

A major issue in achieving effective social distancing has been the movement of people for essential services. This movement causes the gathering of people at places like 'Kirana' stores and vegetable markets and has been a major concern for officials. This movement cannot be completely avoided as people will continue to make movements for activities falling under this category. As the country emerges from the lock-down and restarts its growth engine, social distancing will become a norm and things cannot go back to normal right away. Crowding of people at market-places needs to be avoided at all costs and there is an urgent need to regulate the number of people at public places.

How can we solve this?

We think the gatherings of people for these essential supplies which stands as a major bottleneck for social distancing can be limited and optimized by distributing the movement of people across the working hours. We propose to limit the number of people falling out of homes by the usage of a regulation technique based on the number of requests received to visit a location (like a store, or a local market). The same solution could be extended to a post-lock-down period where limited restriction could be imposed by controlling the movement of users based on a dynamic traffic dependent time allotment for the movement of people in a given area. The number, frequency and time of movement could be subject to users' needs, urgency, the sensitivity f the location under consideration and also possibly users' past exposure (first, second or third level) to positive tested cases.

The Solution

The proposed solution builds on the concept of storing the number of people visiting a local shop as a request in a database, which can then be used to allow a time slot for movement based on previously registered entries. It will minimize possible contact among people by regulating the footfall in marketplaces through a mobile application on which people and vendors need to register. The same app will also be used to send the order beforehand so that the order can be made ready before the user reaches the store, thus effectively reducing the time spent at the store/market. 

Steps in the Proposed Solution

1. Registering on the App - An app supporting regional languages on Android and iOS will have to be deployed for this task. The app will have two fronts- one for the general user, and the other for vendors. The user will have to register himself on the app with a phone number, which will be verified using a one-time-password. The phone number is used to associate each instance of the app to a single user so that it cannot be misused. Once we have the user's address, we can find vendors registered on the app in proximity to the user. The vendors register on a vendor side app where they receive the order when the users select the vendor. 
2. Placing of order - An immediate visit to stores will be allowed only for necessary items like dairy, vegetables, and medicines. For such requests, the user just has to select the store or the vegetable market he/she needs to go to. A time-bound ticket will be generated if that store has the capacity to accommodate more people than the existing tickets for it in the same time period. For scheduled visits, the user can choose a store and register for a corresponding time slot. The user can then proceed to send a list of items he/she requires to the vendor through the app. The vendor can check the availability of each item, and further correspond with the user over an interface on the app. Once the order is finalized the total amount will be sent to the user for payment if he/she chooses to do it in advance. The user can also physically order items at the store in the scheduled slot, but it will be discouraged as it increases the exposure time.
3. Allotment of the slot - The user will be allotted the registered slot based on an algorithm that accounts for the maximum capacity of the store in accordance with social distancing norms, among other factors. Each store will have a maximum number of people that it can serve in a slot, maintaining social distancing effectively. The initial estimate of this number is obtained from the vendor and can be refined at a later stage if it results in crowding. If the slot booking for that store is unsuccessful, then we suggest other stores that can service the user or other slots in the same store which can accommodate him/her. Further information like ratings and reviews of stores can be shown to the user using Google APIs.
4. Collection of order - The user has to collect the order within his/her slot. The vendor will have prepared the order and kept it ready before the slot to ensure minimum contact. To ascertain whether the package has been collected in time, the user will be required to scan a QR code generated on the vendor's side. A special time-bound receipt will also be generated on the user's phone which can be shown to the police.

Technical Details

The infrastructure of an app can be segmented into three parts crudely- frontend, backend, and databases. For android, the front-end will be created using Java on Android Studio. Similar languages and platforms exist for iOS development. The backend and databases will remain the same for both. The deployment of this app will be highly modular - that is, it can be deployed district-wise independently. So the need to suddenly scale-up will translate to creating distinct database instances for different districts, keeping the code base and backend structure the same. The backend and databases will be hosted on a server or a cloud service. Databases are used to store essential data on the server in a secure manner, and backend exposes that data to the app on a request basis. The user addresses are not stored on the app servers but remain only on the user's phone. MySQL databases can be used to store data centrally on the app server. The app will require database tables for storing user and vendor phone numbers, vendor names and addresses, slot bookings for each store, passes generated, user order history and more. The backend can be built upon PHP following the LAMP stack. The effort will be to try and make the app as lean and light-weight as possible, without compromising the privacy of the user. This means that we store data and do computations on the server as long as it does not involve user data. Some of the algorithms which will be executed on the server include finding nearby vendors, assignment of slots in case several bookings are made for the same and suggesting alternative vendors if the slot booking is unsuccessful for a user.

Mechanism for Allotting Slots

In the case of several slot booking requests, we will add them to a request queue. Then we will prioritize the queue based on factors like date of a previous purchase of the customer, distance from the store, recency of order and internal rating of the user. The internal rating is curated using a host of parameters that basically revolve around the user following social distancing. To start with, everyone is given a maximum rating and then penalized if they flout pass rules or do not collect their package in time. This way the slot allotment procedure will not only follow FIFO (First-In, First-Out) method and we will be able to monitor the fairness of allotments to ensure that the maximum number of people get slots.

Feature Phone Compliance

With almost 60% of the urban population owning a smart phone in India, the app will cater to majority of the city dwellers, owing to the fact that it can be utilized by the whole family even if one member posses a smartphone. But the target of actualizing social distancing will fall short if we do not involve the entire population, especially the deprived. To this end, we can introduce a messaging service or IVRS which will respond with a message of 'Yes' or 'No' to whether the phone holder can visit the nearby store or vegetable market. This service can be utilized by any user by simply messaging a keyword to a number specially established for this purpose. The return message will decide whether the user can take the trip or not depending upon the number of requests received in that slot. A similar initiative named "Aarogya Setu IVRS" was launched in association with the Tamil Nadu government. The service was activated once the user gave a missed call to a number and identified his/her location, and provided all assistance, including healthcare and guidance.

On-boarding people and vendors

On-boarding exercise will be incomplete without the support of district administration. The usage of the app or the service will be enforced through the police who will check for the pass/ticket on the user's phone. Further, the stores will be asked to refrain from selling items to customers without the pass. Naturally, there will be difficulties to get the poor to adapt to this system, but it is important to realize that even if the majority of the population is using the system, staggering of footfall to marketplaces will certainly bolster social distancing measures, without restraining the economic activities to certain hours of the day. Moreover, appropriate areas can be identified where people are likely to be less receptive to the app. In such areas, people without passes could be allowed for a specific time, with social distancing enforced through the police, as is the case today. After that period, people with passes could be allowed to come in their slots to the market, which they will prefer over the general period to avoid exposure to the virus.

Expanding to other Areas

The same concept can be expanded to other movements that pose a threat of direct or indirect contact. Some of the areas of application include -

1. Public transport - Gathering at bus/metro station can be limited by issuing limited passes at any given time slot to minimize overcrowding of places, also to ensure appropriate social distancing is maintained in buses and trains
2. Other Shops / Restaurants / Places - A similar permission ticket issuing mechanism can be extended to all possible locations that see a gathering of people. These locations can be classified into zones that can be done from previous data of the traffic they have been getting.

Geofencing

Geo-Fences are of tremendous utility to monitor crowding in demarcated areas. A Geo-fence is a virtual fence or a perimeter around a physical location. Geo-fencing is a tool to connect the virtual experience with the offline world’s physical location. Google uses Geo-fences to tailor advertisements delivered to you depending upon your fine location. Here we use Geo-fences to ensure that market-places do not get crowded beyond a certain threshold, and vary our regulation on passes to that area if such an event occurs. Geo-fences will be most effective for the following:

• Monitoring the crowding at market-places: A Geo-fence can be created over a group of nearby stores. This will help us to register the number of users who are present in it. The minimum radius of the Geo-fence through Google API's should be set between 100 - 150 meters, which will fit our requirements. Once the number of people present inside the Geo-fence exceeds the set limit, we regulate the number of passes approved for stores in that area.
• Preventing misuse of passes: Interaction of the user with various Geofences can help us ascertain if he/she is flouting the pass rules. If the user moves out of a large Geo-fence over the user's house and store location, then the pass can be invalidated. If the user does not enter the Geo-fence over the store location during his slot timings, then his/her internal rating could be reduced. This internal rating will help in competitive situations involving allotment of slots, further incentivizing the user to follow the norms.
• Monitoring of unregulated places: Vegetable vendors are often cramped together and will not possess smart-phones to stagger buyers to different times of the day, through this app. In these cases, officials can identify such places and they can be consequently Geo-fenced. This will also help us to identify vegetable/fruit markets, regulate and monitor the number of people present in them.

The creation of Geo-fences requires fine-location permission from the user, but the app never tracks the user's movements. Only events triggered when the user crosses defined Geofences are known to the app by Google API.

Additional Improvement - Linking with Arogya Setu App

The government of India recently introduced the 'Arogya Setu' app which aids in location tracing by identifying people who might have come in proximity of a COVID+ person using location and Bluetooth. It'll certainly help to curb the spread of the virus if the proposed app can be connected to Arogya Setu, to refrain from people with possible secondary/tertiary exposures from going into crowded spaces. Arogya Setu develops a social graph of interactions and can find out people with possible exposures to a COVID+ patient. If that data can be used to alter the travel movements of these people, then we can minimize the number of people they come in contact with. Arogya Setu app need not share the location history details of these individuals allaying the privacy infringement fears. All that will be required to implement this feature is an API that will reveal if the person has been exposed to a COVID+ person at some level. 

The solution has also been explained in the following video -

Contact Us

We are a team of students from IIT Madras, our contact details are

• Aniket Kamthe ([email protected])
• Aniket More ([email protected])
• Omkar Patil ([email protected])

We are aware that our solution is not perfect, we would love your feedback, please connect with us if you have any suggestion or can help improve and implement this solution.

Edit 1 - Added section of feature phones and vendor onboarding

References