Contact tracing technologies and protocol

May 29, 2020

Content summary

Explaining TAD’s « Contact Tracing Technologies » chart

PURPOSE:

To provide a brief overview of TAD’s Contact Tracing Technologies excel file.

ISSUE:

The contact tracing excel spreadsheet compiled by TAD is a complex, multi-tab spreadsheet, which needs further explanation.

BACKGROUND:

In an effort to keep up with the fast-changing technologies, protocols, and applications being used in digital contact tracing, TAD produced a multi-tab excel spreadsheet.

Contact tracing technologies tab

The first tab focuses on contact tracing technologies by describing the accuracy, tracing method, purpose, and privacy dimensions of three digital contact tracing technologies: Bluetooth, GPS, and Bluetooth-GPS combined. This tab also identifies the differences between location tracking (GPS) vs proximity tracing (Bluetooth).

Contact tracing Protocols tab

Focuses on contact tracing protocols, and describes their underlying technology, relationship with consent, contact matching process (centralized vs decentralized), use of rotating unique identifiers, data retention periods, and safeguards, amongst other features. Each row is accompanied by a brief description of its content. This highly detailed content is provided for the following protocols: Apple/Google, PEPP-PT, DP-3T, BlueTrace/OpenTrace, TCN, MIT SafePaths, and PACT.

Provincial Applications tab

Provides further detail on Canadian apps that have been developed thus far. The only public app adopted by a provincial health authority is ABTraceTogether; however, the Commissioner also requested that MILA’s proposed app be added to this file. Both apps are shown beside each other. The tab describes each app, where it is being used, when it was launched, which protocol it uses, who created it, who uses it, and what information is required to use it.

For additional definitions: please review the Contact Tracing Backgrounder briefing note and the INDU Q&A documentation.

Contact tracing technologies

 

Technologies	Accuracy	Tracing method	Purpose	Privacy
Bluetooth Low Energy	Accurate	Determines user interactions, Based on Bluetooth beacon exchanges and signal strength	Contact Tracing (proximity)	In general, there is no personal identifiable information disclosed. The information collected is usually anonymized using cryptographic algorithms.
Bluetooth Low Energy + GPS	Accurate	Determines user interactions Based on Bluetooth beacon exchanges and signal strength and determines user location using GPS coordinates	Contact Tracing (proximity) + Location Tracking	Privacy can be impacted when aggregating geolocation information.
Geo-tracking using GPS	Less accurate	Determines user location using GPS coordinates	Location Tracking More suited for Quarantine Compliance	Privacy can be impacted when aggregating geolocation information.

Location tracking vs Contact Tracing (Proximity Tracing)

Location tracking—using GPS and cell site information; location tracking is not well-suited to contact tracing because it will not reliably reveal the close physical interactions that experts say are likely to spread the disease. Instead, developers are rapidly coalescing around applications for Proximity tracing, which measures Bluetooth signal strength to determine if two smartphones were close enough together for their users to transmit the virus.

The Challenge of Proximity Apps For COVID-19 Contact Tracing

Contact tracing protocols

Contact Tracing Protocols (CTP)Footnote *. The content of this chart has been solely extracted from the technical documentation that the developers of these protocols have published on their official sites. TAD did not conduct any testing or analysis of these protocols.
Criteria	Descriptions	Apple/Google	PEPP-PT	DP-3T	BlueTrace/ OpenTrace	TCN	MIT SafePaths	PACT: Private Automated Contact Tracing
Sources (Links)	The information sources include the links to the technical documentations published by the developers of the Protocols.
Underlying technology	Contact tracing technology	Bluetooth LE	Bluetooth LE	Bluetooth LE	Bluetooth LE	Bluetooth LE	GPS	Bluetooth LE
Consent is required when uploading records	Interaction records will only be uploaded to the servers if the user consents.	YES	YES	YES	YES	YES	YES	YES
Centralized CTP	Centralized CTP uses a central server that manages, processes, and stores information which can increase the risk of vulnerability to malicious actors by having to breach only the central server to gain access to information.	NO	Centralized contact tracing capabilities	NO	Centralized contact tracing capabilities	NO	NO	NO
Decentralized CTP	Decentralized CTP allows devices to manage and control information independently and allows individuals to pool the information when they want/need thereby decreasing the risk of vulnerability and information accessible to a malicious actor. Decentralized CTPs are considered to be a privacy enhancing solution since there is minimum collection of information.	YES	Decentralized proximity data collection	YES	Decentralized proximity data collection	YES	YES	YES
Rotation of Bluetooth identifiers and proximity IDs	The rotation of Bluetooth identifiers used for tracing interactions can be considered a privacy enhancing solution.	YES	YES	YES	YES	YES	NO	YES
Frequency of proximity ID rotation	The frequency of rotation of the identifiers can be challenging for third parties attempting to track of users.	Every 15 minutes	Not specified	Not specified	Every 15 minutes	Not specified	GPS logged every 5 minutes	Not specified
Data retention period on device	Duration of time which the collected proximity IDs are retained on the mobile device.	14 Days	21 Days	14 Days	Not specified	Not specified	28 days	Not Specified
Data retention period on server	Duration of time that the server retains the interaction identifiers after users that have tested positive upload the interactions IDs. This also aims to identify the data retention time on centralized solutions that collect all user interactions on the back-end server.	Not specified	Not specified	14 Days	21 Days	Not specified	Not specified	Not specified
Contact matching on server	The contact matching process is handled by the back-end server using a randomly generated keys/tokens to compare and cross match to find potential exposed users to send notifications in event of contact with a positive tested user.	NO	YES	NO	YES	NO	NO	YES
Contact matching on device	The contact matching process is managed by the mobile device using a randomly generated key/token as a handshake with other devices to create a list in case of contact with infected user to create notification.	YES	NO	YES	NO	YES	YES	NO
Encryption at rest	A design to prevent unknown entities from accessing, modifying or stealing data on device. Prevents data visibility.	YES	Not specified	YES	YES	Not specified	Not specified	Not specified
Encryption in transit	Used to protect data during communication from device/server or server/device. Encrypts the data before sending, authenticating at endpoints, decrypting and verifying on arrival.	YES (SSL/TLS)	YES (TLS)	YES (TLS)	YES (TLS)	Not specified	YES (TLS)	Not specified
Geolocation tracking	The ability to approximate position of particular devices on a mapping system. This is possible by using GPS coordinates.	NO	NO	NO	NO	NO	YES	NO
Privacy from third parties	The CTP is designed to provide privacy from third parties passively monitoring Bluetooth interaction, in order to identify or track individuals.	YES	YES	YES	YES	YES	YES
Positive results are anonymous to users	Positive results are not known unless user has a PIN from a health authority to send to compare potential users that have come in contact and are at high-risk.	YES	YES	YES	YES	YES	UNCLEAR	YES
Exclusive for Covid-19 contact tracing	The protocol is exclusively to be used with the intent to help minimize the spread of Covid-19 and will be dismantled after the pandemic is no longer a threat.	The system will be dismantled after the pandemic	Not specified	The system will be dismantled after the pandemic	Not specified	Not specified	There is a COVID-specific version, but this project existed before.	The system will be in a dormant state but can be redeployed for future pandemics
Personal information collected	Any information that is not an anonymous unique ID that could potentially identify an individual.	NO	NO	NO	Cellphone Number	NO	NO	NO
Cryptographically anonymized identifiers	Cryptographic encryption and/or hashing algorithms are implemented to prevent the identification of individuals. Link: CSE guidance for Cryptographic Algorithms	The system anonymizes IDs using cryptographic hashes	The system anonymizes IDs using encryption (AES)	The system anonymizes IDs using cryptographic hashes HMAC-SHA256	The system anonymizes IDs using encryption (AES-256-GCM)	The system anonymizes IDs using cryptographic hashes (SHA256)	The system hashes the GPS traces (SHA256), encrypts them (AES-256) and stores the data locally.	Not Specified
Quarantine compliance	Upon a positive result it adds the ability to ensure user is complying with quarantine directives.	Not implemented	Not implemented	Not implemented	Not implemented	Not implemented	NO	Not Specified
Accessed by Public Health Authority	Health Authorities are the only group with the ability to access information.	YES	YES	YES	YES	YES	NO	Not Specified
Development Stage		Phase 1 Launched May 20, 2020. Phase 2 still in development	Still in Development	Still in Development	Launched	Still in Development	Still in Development	Still in Development
Adoption	Jurisdictions that have adopted the protocols or have expressed interest in adopting the protocols. Note: As of today (May 20, 2020), this information is accurate and up to date. Please keep in mind that jurisdictions views have been changing lately, as protocols evolve.	Germany, Italy, Netherlands	Not Specified	Switzerland, Austria, Estonia, Finland and Germany	Singapore, Australia, Czech Republic, Canada (Alberta)	Not Specified	Not Specified	Not Specified
Developed By	Organization or Group that collaborated to create the protocol.	Google, Apple Inc.	Fraunhofer Institute for Telecommunications, Robert Koch Institute, Technical University of Berlin, TU Dresden, University of Erfut, Vodafone Germany, French Institute for Research in Computer Science and Automation (inria)	EPFL, ETHZ, KU Leuven, TU Delft, University of Oxford, University of Torino/ISI Foundation	Singapore Government Digital Services	CovidWatch, CoEpi, ITO, Commons Project, Zcash Foundation, Openmined	MIT, Harvard, The Mayo Clinic, TripleBlind, EyeNetra, Ernst & Young and Link Ventures	Massachusetts Institute of Technology, ACLU, Brown University, Weizmann Institute, Thinking Cybersecurity, Boston University
Additional comments					The developers claim that it is possible to implement other solutions to avoid the collection of cellphone numbers.		There are three versions of this protocol; the data in this column is based on documentation about the third (beta) version, but might quickly become out of date as the projeect changes. This is the foundational protocol for MILA's proposed app. MILA's app adds a "risk model" feature, which adds complexity to the architecture.	Proposed to integrate with Apple My Find. Adding an extension to My Find will take the existing chirps fro contact tracing and will not be encrypted.

Provincial applications

Provincial applications

	Description	AB TraceTogether	MILA	COVID Shield - Shopify	Pandemik - Mimik
Description of Application	The design of the application and what it's intended purpose is.	Designed to help with recording contact tracing connected to positive cases of the coronavirus.	This is a privacy-preserving, location-tracing app spear-headed by academia at MILA. In addition to location-tracing, the app is meant to help users assess their risk of COVID-19 infection and provide them with recommendations to change their behaviour.[REDACTED]	COVID Shield consist of a 3-tier platform that includes a mobile app, a backend server and a web-based portal (optional). COVID Shield comes with the ability to be distributed and managed at a federal, provincial or territorial level and can be configured to each situations guidelines. In which ever capacity COVID Shield is deployed, it will still be able to connect to a centralized backend server.	Pandimik is a mobile application built on Mimik’s hybrid cloud computing platform. The application tracks and records contact history on mobile devices, without storing information on any public server. Pandimik alerts users when it detects that a user may have been exposed to Covid-19.
Deployment	The province the application will be used in.	Alberta	Quebec	Provincial, Territorial and/or Federal	Provincial, Territorial and/or Federal
Status	Is the application available for public use or is it still under development.	Launched May 1	In-development (ongoing)	In-development; Waiting for sponsorship with Public Health Authority	In-development; Waiting for sponsorship with Public Health Authority
Protocol Used	See Contract Tracing Protocols.	Bluetrace/Opentrace	[REDACTED]	Apple/Google	Proprietary protocol
Actors	People or groups that will have access to information.	Government of Alberta	[REDACTED]	Not specified	Not specified
Information Required	When installing the application does the app ask for specific information.	Phone Number	[REDACTED]	No information required, only access to technologies needed for the Apple/Google Exposure Notification Technology	Mimik indicates that after installing the app, users are required to create an account. Nevertheless, no reference has been provided regarding the information required during the account creation process.
Users	Who is the intended user-base.	Albertans	Canadians [REDACTED]	Canadians	Canadians
Sources for information	The information sources include the links to the technical documentations published by the developers of the Protocols.	Alberta.ca	MILA	COVIDShield	Mimik
Additional Comments