A  satellite image of the Yongbyon Nuclear Scientific Research Center in North Korea. GeoEye via AFP
A satellite image of the Yongbyon Nuclear Scientific Research Center in North Korea. Photo: AFP/GeoEye

The “complete denuclearization” of the June 12 Declaration remains ambitious and vague, leaving room for interpretation on both sides. The success of bilateral US-North Korea talks will depend on each party’s definition of extensive disarmament.

Washington is looking for a comprehensive “all-in-one” approach of complete, verifiable, and irreversible denuclearization (CVID), whereas North Korea reportedly is looking for incremental, synchronized nuclear disarmament.

Previous agreements have fallen through in the stages of implementation. To take the steps toward the long-term goal of complete denuclearization, the administration of US President Donald Trump will have to design a long-term balanced framework where both parties are comfortable with the level of transparency and security in a verifiable dismantlement process.

The goal of a verified dismantlement is to obtain high confidence that the program no longer exists and that reconstitution would be difficult and likely to be detected quickly or at least long before significant quantities of banned items are produced. In this sense, the dismantlement is called irreversible.

There exists a need to explore methodologies and synergisms of technologies available to establish a baseline for what it means to have complete denuclearization ensuring security, transparency, and irreversibility to accommodate the political climate and North Korea’s extensive nuclear complex.

Data sources

 The Comprehensive Nuclear Test Ban Treaty Organization (CTBTO), a central authority in nuclear-proliferation monitoring, immediately called for North Korea to sign and ratify the 1996 Comprehensive Nuclear Test Ban Treaty (CTBT) as “part of any comprehensive, negotiated agreement to end the nuclear crisis and achieve denuclearization” which was an essential step toward the “ultimate goal: a world free from nuclear weapons.”

The CTBTO collects, processes, and analyzes data from the 337 facilities of the International Monitoring System, such as the Global Alarm System to detect all nuclear explosions and 80 extremely sensitive radionuclide sensors to detect radioactivity. It also registers dispersion of radioactivity stemming from all other sources anywhere in the world, in particular to nuclear accidents.

Verification – measurements and procedures that the negotiated activity is taking place – is often done through the traditional National Technical Means systems, but there is a serious need to utilize the full range of technologies to detect and characterize nuclear activity, equipment, and materials in substantive, verifiable, and secure ways.

Open-source data, seismic signals, and geospatial information analysis are used as well as other mediums to mitigate challenges in detecting nuclear activities. A hypothesized method on the horizon is Public Technical Means, one where concerned citizens and the public participate in the process of arms control using smartphones or citizen-adapted sensors to collect and analyze data and information.

Social media are also cited as tools, to examine the field verification activities. However, this methodology also must accompany a larger public campaign about data privacy and mitigate potential deception and falsification, and in a country like North Korea, where ICT (information and communications technology) infrastructure is seriously constrained, the option to use public technical means to monitor verifiable denuclearization likely is not available.

There are audio-visual tools to evaluate the completeness and correctness of information. Multimedia data, such as photographs, videos, and audio recordings at nuclear facilities, may be used to enhance preparation for onsite verification by helping to visualize physical attributes of facilities and equipment, thus aiding in such safeguards activities as design information verification and environmental sampling.

Another emerging source is Earth observation tools such as satellite imagery equipped with high-definition videos, and synthetic aperture radar systems, and thermal infrared systems. The deployment of HD video cameras on small satellites could facilitate close monitoring of motion, including short-lived or quickly changing phenomena, to detect fissile materials.

Data fusion model

Voluminous amounts of data are being generated on a daily basis from multiple data streams. The most effective way to monitor verification of nuclear facilities is for the data to proceed and stored in real time. The heterogeneous data streams can be integrated through validation algorithms and systems.

We can detect anomalies in the data, such as unusual nuclear activities or potential illicit activities in patterns of behavior or what constitutes nuclear weapons development activity prohibited under the treaty, drawing a line between activities exclusively weapons-oriented and those peaceful or for dual use – serving military and/or civilian purposes. This also would help clarify whether an application other than nuclear weapons development is plausible, such as a firewall to look at the presence and absence of activities and their evolution over time.

There exist many approaches today, but one most apt for the case explored is efficiently fusing heterogeneous data streams by gathering a network of dependencies within a dataset and compare the different networks to quantify the similarities. A mathematical proof-based deductive process can be hard to scale because of the exact modeling of the programming language; therefore, the best option is to fuse an algorithm to create a normative pattern.

First of all, the data has to be collected in a standardized matter and cleaned up so each dataset contains similar fields. For the data to be digestible and interpretable, the model can use an application programming interface or a systems integration application layer that can interface with different systems into the data fusion model.

Data fusion can be modeled with different methods and techniques based on the relationships of the various data sources. For instance, two sensors monitoring the same site could be considered as redundant to increment the confidence of the data. But if the signals are from two different viewpoints, they could be complementary. When fusing new information with previous data, such as additional audiovisual tools, it would be considered as cooperative.

This pattern-based vulnerability discovery graph would in essence be a data governance model, acting as a meta-data layer as a single version of the truth. It acts as a data-validation layer providing integration access through the data fusion running the anomaly detection software among distributed sources.

As participation allows, using novel open data, multimedia, or geospatial data can be used as synthetic data to ensure we are able to detect the anomalies with minimal false positives. Later, injecting real data into the dataset can validate the algorithm for no false positives.

The data fusion model and its overlapping interface would allow analysts to detect unusual patterns of events and activities that support the attribution of the source of illicit activity.

Blockchain, machine learning

US officials worry that North Korea could hide key nuclear materials or facilities. Now, how can we ensure no entity purposefully tampers with the data?

Blockchain’s applications, such as storing records, clearing and settling accounts, and ensuring validity and execution of contractual arrangement, could allow us to see when data has been viewed or modified. It would ensure data integrity in nuclear or satellite command and control to log crucial information permanently on whether a hacker has modified something in a database.

Controlling the spread of nuclear-weapons technology requires a rainbow of decentralized, sometimes overlapping and sometimes fragmented systems of international agreements, informal arrangements, and national legislation. Not surprisingly, differences in national implementation and enforcement continue to frustrate efforts to keep dual-use goods and technologies out of proliferator hands. These implementation gaps, coupled with the sheer volume of global trade and commerce, have reduced the barriers to entry for intermediaries and created pathways for illicit procurement networks to exploit.

Blockchain has been discussed as a method in the supply-side controls, but there remains discussion on whether blockchain-based systems will help or hinder monitoring for illicit transactions and fraud. It opens up possibilities for transactional monitoring, as it could help ensure transparent and real-time monitoring across all transactions. Moreover, trade and finance systems based on blockchain mean that financial transactions need not be conceptually separate from trade-related transactions (such as shipping and insurance), thereby increasing the granularity of information available to regulators, enforcement, and intelligence agencies.

Any measurement system in such a regime necessarily requires tamper-proof certification and authentication. As is easily imagined, the security system is what assures the integrity of the process to prevent both cheating or leakage of military secrets. The North Korean leadership likely suspects the other parties’ commitment to rewards and their motivations for conducting certain verification activities. This dismantlement model recognizes the initial lack of trust among the parties and allows confidence-building through the successful implementation of initial cooperative steps toward verified dismantlement.

Verification would occur throughout the process, and incentives and security guarantees would be implemented in stages linked to the dismantlement and verification steps.

Artificial intelligence (AI) is concerned with building algorithms that are capable of working with (processing, or operating with) data while it is still in an encrypted state. As any part of a data process that involves exposing unencrypted data represents a security risk, reducing these risky incidents could help make things much safer. The AI data fusion model could help make the decision about whether certain transactions should be investigated. Having each decision recorded on a blockchain to be audited with the various stakeholders would ultimately be a machine-learning-powered algorithm with a decision-making process.

The way we could achieve this is by integrating the data fusion model into an encrypted ledger. Segmented verified databases built and implemented only upon databases that have been verified can reduce the stress of troubleshooting and finding abnormal datasets.

The data quality assurance or maintenance of the server can be offloaded on to a cloud, potentially maintained by the CTBTO to have a third party in which all stakeholders are politically comfortable.

The blockchain model would be in a permissioned ledger to ensure only a few selected members could access it, and distribute the files across the network to have files available even if part of the network is broken. This would ensure the file would not be tampered with without detection, a highly robust database that can only be read and updated only with permission.

When data has been created for the purpose of model building, licenses that cause restriction or permission can be specified by the party. Blockchain technology then makes the process of doing this relatively easier.

A centralized or permissionable blockchain model, where roles are granted to certain users to interact with the ledger for a trusted set of users, would only allow for the International Atomic Energy Agency member states or selected parties to view the data. For instance, when a report is transmitted into the data, it would be distributed across separate entities. They would be able to view when and if the data is valid, unique and authenticated, as verified by the data fusion model. Using private key encryption, the IAEA could also only share the metadata deemed suitable for sharing, and only allow certain entities with permission to view more sensitive information.

The Trump administration is now looking for ways to continue peaceful talks to sustain the momentum to pursue a creative, realistic, and conciliatory way to move forward with credible steps toward disarmament of North Korea. The task will be accomplished by states willing to make the necessary political commitment and resources, with detailed models and verifiable dismantlement of gas centrifuge enrichment and nuclear weaponization programs for application in North Korea.

Grace Kim has published research on human rights, cultural anthropology, and innovative philanthropy. With experience at the Korea Desk and Policy Planning Staff at the Secretary's Office of the U.S. Department of State, she has provided support to strategic guidance to U.S. foreign-policy priorities.

One reply on “Leveraging technology for a peaceful Korean Peninsula  ”

Comments are closed.