AI-Powered Monitoring & Analysis of Bitcoin Transaction Traffic
National Technical Research Organisation (NTRO) · Transportation & Logistics · Software
The dataset is provided and the specification is complete, so this is genuinely buildable — make the network-plus-blockchain correlation your distinctive angle, and treat the explainability as core since a lead an investigator cannot understand is useless.
What it actually is
Bitcoin's pseudonymity lets criminals move and launder illicit funds while evading normal financial surveillance. The ask is an offline system that ingests bulk Bitcoin transaction and network metadata, links network-layer observations like IP and timing with blockchain-layer data like wallets and amounts, and applies machine learning to flag anomalies, cluster entities and produce ranked, explainable investigative leads.
What to build
An offline analysis system for Linux that ingests bulk metadata records containing timestamp, source and destination IP and port, transaction ID, input and output wallet addresses, amounts, fee and script type, builds an entity-transaction graph linking IPs, wallets and transactions, applies a trained machine learning model — not just rules — to detect anomalies and cluster addresses likely controlled by the same entity, and produces a ranked, explainable alert list stating why each wallet or transaction was flagged with a confidence score, presented through a link-analysis visualisation; the description confirms the dataset will be synthetic, modelled on real Bitcoin fields.
Smallest thing that wins the room
Load the synthetic transaction dataset, build the entity graph, and show the system clustering several addresses into one suspected entity and flagging a layering pattern, with the top-ranked alert explaining in plain terms why it was flagged and the link-analysis view showing the fund flow.
How crowded this one gets
A guess, projected from the 2025 statements — the last year where both the submission counts and the winners were published.
Quieter than 15% of the 226 · #192 of 226 by expected field · reaches the 500 cap
Busier than most. Expect several teams to arrive at the same obvious solution.
Why: defence, intelligence and space bodies drew small fields.
This is a guess, not a fact
Nobody has published 2026’s numbers yet. This is an analysed estimate from last year’s pattern, so please do not take it as the truth — check the live counter on the SIH portal before you decide anything. The range covers the middle half of likely outcomes, so one statement in two lands outside it. Entry closes at 500 ideas per statement, so no range goes past that — a statement that reaches the cap fills and shuts rather than drawing an unlimited crowd. The model reads only three things a team can see before choosing — software or hardware, the theme, and what kind of body posted it — and those explain about a quarter of the variation in last year’s field sizes (R² 0.25 on held-out statements). Trust the band more than the number, and the ordering more than either. It cannot see how good your idea is, which is the part that actually decides it.
The scores
The number is the shorthand. The line under it is the reason.
Acceptance potential
4/5A strong pick — the dataset is provided, the specification is complete, and the network-plus-blockchain correlation is a genuinely distinctive angle that separates this from the many purely on-chain analysis projects, with NTRO's investigative framing giving it clear purpose.
Feasibility
4/5The description states a synthetic dataset will be provided modelled on real Bitcoin fields, address clustering heuristics are well-documented, graph analysis and anomaly detection are standard, and everything runs offline on Linux with no live blockchain access needed.
Innovation scope
4/5How you fuse the network-layer IP and timing data with the blockchain-layer wallet graph is the genuinely open part — most public blockchain analysis uses only on-chain data, so correlating it with network metadata is a real and distinctive contribution.
Clarity
5/5The description lists the exact input fields, the required graph construction, mandates a trained model rather than rules, specifies ranked explainable alerts with confidence, and even defines the synthetic dataset schema — leaving nothing ambiguous.
Effort
HeavyIngestion, entity graph construction, the ML detection model, explainability and a link-analysis visualisation are five connected pieces, though each rests on established techniques.
Demo-ability
EasyA link-analysis graph showing wallets clustering into an entity with a fund-flow trail highlighted is a compelling, investigative-looking demo that reads instantly.
In its favour
- Green flag: The synthetic dataset is provided with a defined schema, so there is no data-sourcing risk and everyone works from the same fields
- Green flag: Correlating network-layer IP and timing with the on-chain wallet graph is a distinctive angle that most blockchain analysis omits entirely
- Green flag: The common-input-ownership clustering heuristic is well-documented, so your entity clustering rests on established forensic technique
- Green flag: Ranked explainable alerts are exactly what a real investigator needs, and the explainability requirement pushes you toward genuinely useful output
Against it
- Red flag: The network-blockchain correlation only exists because the synthetic dataset pairs IPs with transactions — in reality that linkage is rarely available, so acknowledge that your capability depends on the provided data model
- Red flag: Address clustering heuristics produce false merges, and over-clustering wrongly implicates unrelated parties, so cluster precision matters in a forensic context
- Red flag: The description requires a trained model rather than rules, so a purely heuristic submission does not meet the stated requirement
- Red flag: Flagging a wallet as suspicious is an investigative accusation, so explainability and confidence must be genuine rather than decorative
What you will be writing
- NetworkX / Neo4j entity-transaction graph
- Address clustering (common-input-ownership heuristic)
- Graph anomaly detection / GNN entity classification
- SHAP or rule-trace explainability
- GeoIP / ASN enrichment
- Cytoscape.js or Linkurious link-analysis UI
- Blockchain forensics
- Financial crime investigation
- Graph analytics
Prior art to read before you start
cryptocurrency transaction analysis · entity clustering and de-anonymisation · explainable investigative lead generation
Analysed by Claude Opus. Every score above is a judgment call with its reasoning attached — kindly cross-check this against the official statement on the SIH portal before your team commits to it.