Cryptocurrency has become used all around the world and is continuing to grow. Total daily trading volumes of cryptocurrency are said to reach $275 billion per day. But how many of these transactions are actually legitimate? It is estimated that in 2024 alone, there were $51.3 billion worth of illicit transactions on the crypto market, with that number expected to rise in 2025.
CryptoGuard proposes a solution to this – an advanced blockchain monitoring tool to enhance transparency and security in cryptocurrency transactions. Featuring a real-time dashboard, users can track Bitcoin transactions and utilise search functionalities to filter by user. Machine learning algorithms detect suspicious activity, such as ransomware payments and money laundering, and flag transactions for analysis. With its secure, reliable, and interoperable architecture, CryptoGuard is a vital tool for fraud detection.
Name: Alexander Smith
Student number: 47445035
Focus: Provide a working prototype that enables minimal transaction monitoring and search capabilities, while upholding crucial quality attributes.
Dashboard:
Fraud Detection and Monitoring:
Security:
Security is a crucial attribute for this project because it directly impacts the integrity of the data and dependability of the system. A critical component is the ML model, designed to identify and detect illicit activities. This raises significant concern if a threat actor can gain unauthorised access to the dashboard and modify the transactions. With access, a threat actor can manipulate transaction data to evade detection by the model or falsely flag legitimate transactions. As a result, the integrity of the system would be compromised. Implementing strong access controls and audit logs will ensure that only authorised users can interact with the dashboard. Relevant penetration testing attacks can be carried out to test the security of the system, identifying vulnerabilities that require mitigating.
Reliability ensures that the system provides accurate and up-to-date transaction information. Since system administrators rely on the dashboard to track suspicious transactions, somewhat real-time updates. Delays in reporting transactions could allow illicit activities to go unnoticed, reducing the effectiveness of the monitoring model. A robust data pipeline is required to ensure that transactions are consistently retrieved and displayed with minimal latency. Redundancy and error handling should be in place to prevent data loss or incorrect reporting. Comparing the CryptoGuard dashboard with other blockchain dashboards will help ensure that the information provided is accurate and up to date.
With CryptoGuard being built upon multiple components, it is essential that they can communication to exchange information. All transactions must be accessible to the search-and-filter feature, the ML-based fraud detection system, and the alerting mechanisms. Users must be able to efficiently filter through transactions and find relevant records. The ML model requires access to both real-time and historical transactions to detect patterns of suspicious behaviour. Finally, accurate reporting on flagged transactions requires a detailed and up-to-date database. Improving interoperability can be done using effective database schema, which can then be stress-tested with large amounts of transactions.
The ability to scale the project is crucial for its longevity. Initially, the system will focus on Bitcoin transactions as this is used for many illicit payments. To be truly effective, it should support multiple cryptocurrencies. This means the system architecture must be scalable, allowing new cryptocurrencies to be integrated without requiring significant modifications. The ML model should also be designed such that it can detect suspicious behaviours regardless of the currency. Furthermore, as the volume of transactions increases, the system must be capable of handling large datasets effectively, ensuring that performance does not degrade over time. Testing this can be done by adding extra currencies or putting the system under a significant load.
Each of the attributes listed as well as the functionality of the project can be broken down into a quantitative and qualitative evaluation framework.
Ensuring the integrity and confidentiality of the dashboard means carrying out a thorough security evaluation. Verify that only authorised users can alter dashboard settings through access control testing and attempt to make unauthorised changes through simulated penetration testing attacks. Data integrity is another pivotal aspect to consider, which can be tested through SQL injection attacks.
Evaluating the reliability of the dashboard is important to ensure accurate and timely data transactions. One key aspect to test will be the time taken to display new transactions, which should be < 10 seconds. Measure the frequency of failed transaction retrievals to calculate an error rate, as this will be indicative of the reliability of the project. Finally, simulate varying loads of user traffic and transaction traffic to ensure stability.
Interoperability can be evaluated by thoroughly testing the MVP itself. Verify that the search feature correctly filters through up-to-date transactions based on a user input. Confirm that the model can access real-time and historical transactions for accurate flagging and reporting. All transactions – including flagged ones – must be stored in such a way that it is accessible to the model and all parts of the dashboard.
Future advancements of the project are made viable if the system is scalable. Therefore, the system must uphold an acceptable transactions throughput, handling at least 7 transactions/s. Additionally, the database must be able to handle growing numbers of transactions, still maintaining an efficient query speed. With both of these implemented, CryptoGuard should be able to handle multi-currency support depending on the generalisation of the ML model.