Original Title：Hacking Bluetooth for COVID-19 contact tracing

　　Traditional manual contact tracing is not fast enough to keep pace with the rapid spread of COVID-19. Here, technology can shorten the delay between confirming an infected case and isolating all contacts.

　　Among the countries that have managed to get the virus under control, South Korea and Germany are using digital tracing to contain and suppress the epidemic. While this offers hope, it is important to keep in mind that technology is only one tool, albeit an important one, and not a comprehensive end solution.

　　While investing in digital tracing, design engineers must continue to evaluate the benefits against cost and risk. As governments and citizens grow eager to return to a pre-COVID world, it is easy to get too enthusiastic about technology’s potential to save lives and rescue economies.

　　Digital tracing should be incorporated into a traditional contact tracing procedure and not replace it. Tracing interviews conducted by medical professionals are a fundamental process in contact tracing and should be used in tandem with technology. Furthermore, engineers don’t yet have proven, mature tracing technology. Available techniques are error-prone, which poses the risk of false-positives and false negatives. Additionally, as mentioned earlier, there is not sufficient data to legitimately back up claims regarding how this virus is spreading.

　　Hacking Bluetooth for contact tracing

　　Bluetooth Low Energy (LE) is a ubiquitous, widely-deployed, and mature connectivity technology with enormous potential to enable digital tracing. Every new smartphone is equipped with a Bluetooth radio, and almost everyone has a smartphone. If we reprogram smartphone Bluetooth radios in a certain way, we can potentially use smartphones to detect which individuals came in direct contact with each other.

　　From a technical standpoint, the theory is simple. Every smartphone Bluetooth radio would be tuned to broadcast short random messages to nearby phones. The short random messages form an encrypted signature. Simultaneously, every phone would listen to and record all the encrypted signatures they are receiving from nearby devices.

　　Next, each phone would construct a record of all the other phones it came in contact with. By populating an online database, engineers could construct an entire population tracing record. If an individual “A” is diagnosed with COVID-19, an immediate notification can be sent to all contacts, asking them to self-quarantine.

　　Digital tracing merits

　　The Bluetooth-based digital tracing is superior to traditional contact tracing methods in a few ways:

　　Immediacy: Once a case is confirmed, contacts can be notified quickly. Immediacy can be crucial in whether we can slow down this virus or not.

　　Recursive contact tracing: With digital tracing, we can warn people that they may have come in direct contact with a confirmed subject (Figure 1).

　　Resource optimization: It can relieve the stress on medical resources. A notified subject can still take precautionary measures even if testing is not an option.

　　Figure 1 Digital tracing can trigger immediate self-quarantining and decontamination response to confirmed cases and recursive contact tracing. Source: American Association for the Advancement of Science

　　On paper, everything looks good. However, in reality, there are a few serious challenges that must be overcome. First and foremost, Bluetooth was not built with contact tracing in mind. So, there is a need to re-engineer some features of Bluetooth radios to unlock location tracking. For instance, the 2.4-GHz band that Bluetooth operates in is heavily occupied by other radio technologies, such as Wi-Fi. The high volume of traffic on this band can affect the reliability of Bluetooth.

　　However, because Bluetooth is widely deployed and consumes minimal energy at a reasonable cost, it makes the most sense to hack Bluetooth and work around its weak spots. This might be the most effective way to achieve a digital tracing technology within a reasonable timeframe, considering the urgency of the global health crisis.

　　The RSSI technology

　　Received signal strength indicator (RSSI) technology offers the Bluetooth digital tracing solution closest to deployment. It uses information about received signal strength to determine the distance between two Bluetooth radios.

　　Unfortunately, there are a few more factors that play into signal strength other than distance. In outdoor environments, humidity, temperature, and rain can significantly affect signal strength. In indoor environments, objects, multipath reflections, and signal blockers can greatly impact the strength of the signal received.

　　Hence, Bluetooth RSSI distance resolution is limited to 3 to 5 meters, with higher than desirable rates of false negatives and false positives. It means that engineers can only count on this technology to assess whether two individuals have been in each other’s proximity and for how long. A carefully-designed Bluetooth RSSI tracing algorithm can classify exposure into close and far encounters.

　　However, RSSI won’t be able to assess the actual distance between contacts, which would have been valuable data. Regardless of its limitations, RSSI can still enable a paradigm shift in contact tracing.

　　An immediate notification sent to smartphones would trigger precautions from exposed individuals. Yet, given that a false negative could threaten human life, RSSI must work with traditional professional tracing investigations.

　　Bluetooth direction finding

　　An important advantage of RSSI over competing alternatives is its simple implementation, allowing a shorter development cycle for a smartphone app. It means that major technology companies can build a simple API around the Bluetooth radio that developers can use to build their tracing apps. Other location tracking Bluetooth technologies that could achieve more granular location data require access to low-level features of Bluetooth radios, posing severe privacy issues for smartphone users.

　　A major alternative to RSSI is angle-of-arrival (AoA) and angle-of-departure (AoD) solutions, and both fall under the category of Bluetooth direction-finding technologies. Bluetooth direction-finding solutions provide sub-meter location accuracy and are ideal for dense urban areas such as shopping malls, grocery stores, and office buildings. RSSI would have an unreliably high error of false positive and false negative in a dense urban area. On the other hand, Bluetooth direction-finding can provide more granular data of person-to-person contacts.

　　Bluetooth direction-finding location measurements use the same principles of radar and phased-array technology (Figure 2). An antenna array receiving the same wave from a distant source would produce multiple waveforms with phase variation between the different paths, depending upon the direction of the received wave.

　　Figure 2 Angle-of-arrival (AoA) technology can take location detection accuracy to the sub-meter level.

　　By using two antenna arrays—beacons—at known locations in a facility, the location of each Bluetooth tag in a building can be detected within sub-meter accuracy. The trick employs a few geometry techniques to estimate the distance and angle of each tag from the two locators.

　　However, in reality, the situation is a lot more complicated due to multipath, polarization, and jitter variations. This is where sophisticated algorithms step in to correct errors and tighten the accuracy of location measurement to less than a meter.

　　Smartphone app vs. Bluetooth bracelet

　　This article aligned Bluetooth-based digital tracing solutions with medical expertise and presented two technologies to support traditional contact tracing: RSSI and Bluetooth direction-finding. Stay tuned for the second part of this series, which will show how these technologies can be implemented to create a contact tracing system on Bluetooth bracelets.

　　Editor’s Note: This article is for general information purposes only and provides an overview of a specific developing situation that continually evolves. It is not intended to, and should not be construed, for public health guidance. —Majeed Ahmad

　　Asem Elshimi is an RFIC design engineer for IoT wireless solutions at Silicon Labs.