Hala Systems: Blockchain Case Study for Saving Lives and Mitigating War Crimes

Documenting War Crimes in Syria

聽

The Challenge

The Syrian Civil War has claimed the lives of over 500,000 innocent civilians, including 20,000 children and 13,000 women since March of 2011. The U.N. estimates that 3 million more lives will be at risk if Syrian President Bashar al-Assad continues the air campaign against his people. This campaign has had a devastating effect all over Syria, particularly in the province of Idlib, where an estimated 1.5 million civilians displaced from other parts of Syria reside.

Hala鈥檚 Sentry Early Warning System has successfully been used as an airstrike alert system, pinpointing the location and alerting citizens of an impending attack. The system has saved hundreds of lives in Syria, while effectively reducing the lethality of such attacks in heavily bombarded communities by an estimated average of 20 to 27%.

Unfortunately, as Hala is a private enterprise, it technically 鈥渙wns鈥� the data generated by Sentry. As a result, it is possible to claim that any data given to justice and accountability organizations may have been tampered with or mishandled through the reporting process regardless of intent. This makes it difficult for the data to be used in a court of law and prove its credibility to stakeholders.聽

Can blockchain technology be leveraged to immutably store war crime information, and identify perpetrators of attacks on schools and hospitals?聽

聽

The Ethereum Solution:

Decentralization and Data Integrity

Hala Systems contacted ConsenSys to build a solution that sent the signed meta transactions directly to the public Ethereum blockchain to prove that the information gathered has not been tampered with and/or modified in any way between the time and place of its gathering and the present. Additionally, using a distributed ledger database increases the efficiency of the already existing Sentry Early Warning System and develops it into a tool for immediate and accurate reporting of war crimes.聽

By leveraging the Ethereum network, Hala is able to effectively remove itself from the core of the Sentry reporting equation. When a report is submitted, the timestamp and geotag of when and where the threat was detected is hashed to the Ethereum blockchain where it lives forever. On-chain, it can be referenced but not tampered with. If the data is tampered with, the associated hash value will reflect that鈥攅nsuring complete transparency.聽

By automatically submitting and storing data onto the blockchain, Sentry is able to prove that credible data has not been changed, edited, or tampered with in any way from the time and point of collection through to the present.聽

Additionally, by placing metadata on-chain, Hala can avoid common issues with data collection in high-risk areas. For example, If the IoT ground sensors malfunction or falsely detect a non-threat event, there will be no corroborating open-source data such as social media posts or media reporting to reference.聽

Hashing all of the human and machine-sourced data allows for easy organization and identification of threats and attacks by creating a publicly-accessible reference point of time and location. When analyzed, an 鈥榠nsight鈥� is made based on certain parameters and conditions that indicate a credible imminent threat or executed attack. These warnings are then issued to at-risk communities who can use the information to remove themselves from harm鈥檚 way.

Additionally, sending the data points to the blockchain instead of a Sentry server provides immutable data collected from the ground in violent places. This data collected creates a critically important tool for accountability and justice efforts. Such efforts help prevent or mitigate tomorrow鈥檚 violent conflicts.

Why Use Blockchain to Document War Crimes?

The purpose of leveraging blockchain technology in combination with Hala鈥檚 Sentry system is to ensure that data collected by the system is trustworthy. Think of the blockchain as a signaling mechanism. By hashing the raw data (time, geographical location, and report content) reported by the human and machine observers and hardware sensors, we create an immutable reference to an event that likely occurred at a certain time, at a specific location.聽

Could the sensors have malfunctioned, or falsely detected a non-threat event? Absolutely, but in such a case there will be no corroborating open-source reporting, such as social media posts, images, and media reporting. When an attack occurs, there are consistently multiple sources of reporting outside of Hala鈥檚 control. By hashing the event generated by sensors and human observers, a very strong focal point is created around which these publicly sourced data points can be organized. The outcome is an immutable signal written on the public Ethereum blockchain and a number of off-chain data points validating the occurrence of the event (and vice versa). This creates a compelling, and detailed record.聽

By creating SHA256 hashes of event information, Hala鈥檚 system will generate tamper-proof reporting that an event occurred. SHA256 is a cryptographic hash algorithm that provides an (almost) unique, 256-bit (32-byte) output for every input that it is provided. Since the algorithm can generate so many unique outputs (approximately 1.16 x 1077 outputs) it is assumed, in practice, that all inputs create a unique output. When journaling digital information, this means that we can convert normal data, like IoT sensor data and report forms submitted by mobile applications, to unique SHA256 hashes, and, later in the future, take those data sets and compare them back to their original, respective hashes.聽

If a report is tampered with in Hala鈥檚 database after we hash it, the new, manipulated report will not generate the same hash that it created before. Since the original hash of the report does not match the new hash of the report, we know immediately that the report was tampered with.聽

聽

How It Works

Using a combination of civilian observations and social media reporting, IoT sensors, and location, time, and environmental metadata collected by these sources, Hala is able to create and send out a warning in advance of danger to civilian communities. These warnings are useful both to the civilians themselves and to the accountability experts who seek to ensure international humanitarian law is enforced.

1. Observation is made: User or machine observes an event of interest
2. Report is created: User or machine creates a timestamped and geotagged report for submission
3. Report is hashed to blockchain: The timestamp and geotag are sent to the Ethereum blockchain where they exist immutably and accessible forever
4. Report is transmitted: Report is submitted by app or sensor and transmitted to Hala server
5. Report is processed: Collected data is processed further by Hala team

How It Works for Users

Sentry is premised on a simple fact: The more time someone has to prepare for an airstrike, the higher their chance of survival. And now lots of people are relying on Sentry for that edge: 70,000 follow the local Facebook page. Its Telegram channels have 16,400 subscribers. A local radio station broadcasts Sentry alerts鈥攁nd the digital network of witnesses is only growing.聽

聽

Goals Achieved聽

Lives Saved

Analysis indicates that warnings reach up to 2.1 million people, have saved hundreds of lives, prevented thousands of injuries, and reduced traumatic anxiety for hundreds of thousands of people. According to a preliminary assessment by Hala, the system reduced the lethality of airstrikes by an estimated 20-30 percent in several areas under heavy bombardment in 2018.聽

• From local surveys conducted, Hala found that people need a minimum of 1 minute to seek adequate shelter. Sentry now averages a warning time of eight minutes.
• The Hala team was able to fine-tune Sentry鈥檚 predictions to be accurate to within 30 seconds of the war plane鈥檚 arrival.
  
• The initiative has processed over 50K pieces of digital evidence
  
• According to Hala, at optimal range Sentry can now identify threatening aircraft about 95 percent of the time.

Immutable Evidence On-Chain as a Reference, Critical Data Sets, and Provenance Record of War Crimes

• By leveraging ConsenSys-built technology, Hala Systems can now collect data and prove in International Criminal Court that their data integrity is sound and untouched from the time it was collected to the time it鈥檚 used to prosecute alleged war criminals聽
• In addition to the primary value derived from the system, one secondary benefit is that it provides reporting deemed valuable by accountability-focused organizations such as the United Nations Office of the High Commissioner and the International Criminal Court.聽

Hala鈥檚 reporting has been used to augment investigations into human rights abuses, war crimes, and violations of ceasefire agreements.聽

聽

Next Steps

Longer-term, the platform developed as part of this PoC could be scaled as a production system and function as a protocol for automated blockchain reporting (of specified events). A protocol that enables IoT devices to speak to, and record immutable evidence of events & actions on the public Ethereum blockchain alleviate a large portion of the verification issues that currently restrict the scalability of blockchain solutions in the social impact sector. It could be leveraged across nearly all blockchain verticals.

Hala Systems continues to improve their product and life-saving offerings for Syrian citizens. In efforts to reach citizens without cell phones or power, Hala modified a siren by adding a remote-access component. The team shipped 150 prototypes around the country and is currently conducting a way to send alerts to non-mobile users.聽

Hala Systems continues their research to determine which blockchain environment is most viable for the purpose of journaling immutable evidence as they tackle the feature requirements of 鈥榠mmutability鈥� and 鈥榲eracity.鈥� In addition, the Hala team is developing methods to increase the scale of their solution, as well as incorporate usage of IPFS.

聽

About Hala Systems

John Jaeger, a tech expert for the US government, and Dave Levin, founder of Refugee Open Ware teamed up with Syrian developer, code-named 鈥淢urad,鈥� to form an early threat detection and reporting group Hala Systems.聽

Sentry started as a handful of if/then statements, a logic tree, and an Android app. The team gathered related data, e.g., aircraft monitor data, plane type, trajectory, previous flight patterns, lift-off location, cruising speeds, and connected it with reports of where those planes dropped their bombs. Hala Systems plugged the data into a formula that accurately predicted where the warplanes were most likely headed and at what time.

Going for more in-depth and reliable source entry points, the team sourced critical information from fighter pilots who had defected from the Syrian air force. To further fine-tune their predictions, Hala Systems hired a team of Syrians to examine video and social media evidence and cross-reference it against Sentry鈥檚 predictions. To increase diagnostic accuracy, the team created a software that, with the help of a neural network, could search Arabic language media for keywords that helped to confirm the location and timing of an airstrike. More data on more air strikes meant better information and better predictions.聽

After many iterations, the Hala team was able to fine-tune Sentry鈥檚 predictions to accurately denote a war plane鈥檚 arrival within 30 seconds.