💻 computer science

DBHMT: Dynamic Balancing Hybrid Merkle Tree– Design, Optimization and Evaluation in IOTA Tangle for IoT-Enabled Healthcare 4.0

This paper proposes the Dynamic Balancing Hybrid Merkle Tree (DBHMT), an optimized data structure for IOTA Tangle-based Healthcare 4.0 that employs a dynamic pruning mechanism to achieve a fourfold reduction in storage costs while maintaining security for resource-constrained IoT environments.

Original authors: Basker Palaniswamy, Md Noor-A-Rahim, Paolo Palmieri, Francesco Regazzoni, Ashok Kumar Das, Vivekananda Bhat K

Published 2026-07-07

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Basker Palaniswamy, Md Noor-A-Rahim, Paolo Palmieri, Francesco Regazzoni, Ashok Kumar Das, Vivekananda Bhat K

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: A Digital Filing Cabinet for Health Data

Imagine a massive, shared digital filing cabinet used by hospitals and doctors to store patient records. In the future of healthcare (called "Healthcare 4.0"), this cabinet needs to be incredibly secure so no one can tamper with the records, but it also needs to be small enough to fit on the tiny computers inside wearable health devices (like smartwatches or heart monitors).

The problem is that the current way of organizing these records (using a standard "Merkle Tree") is like a filing cabinet that grows super-exponentially. Every time a new patient record is added, the cabinet doesn't just get a little bigger; it explodes in size. For devices with limited storage, this is a disaster.

The authors of this paper propose a new, smarter filing system called DBHMT (Dynamic Balancing Hybrid Merkle Tree). They tested it on a specific type of digital ledger called the IOTA Tangle (a blockchain designed for the Internet of Things).

The Core Problem: The "Growing Tail"

To understand their solution, imagine the patient records are arranged in a tree structure:

The Right Side (Static): These are patients with long-term or lifelong conditions (like diabetes or heart failure). Their records stay in the cabinet forever.
The Left Side (Dynamic): These are patients with short-term issues (like the flu, a minor cut, or a seasonal allergy). Their records are only needed for a short time.

In a standard system, the "Left Side" keeps growing outward, creating a massive, messy "tail" of old, expired records that no one needs anymore. This clogs up the storage space.

The Solution: The "Dynamic Pruning" Gardener

The authors' solution is to act like a gardener who knows exactly which branches to cut.

The Hybrid Tree: They built a tree that is part "n-ary" (many branches) and part "binary" (two branches), designed specifically to handle the mix of long-term and short-term patients.
The Pruning Mechanism: They introduced a "pruning factor" (let's call it $\alpha$ ). This is a rule that says, "If a patient's record is expired (like a cold that has healed), cut off that branch of the tree."
The Balance: The tricky part is that if you cut too much, you lose security (you can't prove the record existed). If you cut too little, you run out of storage.

The "Goldilocks" Search (The Optimization)

Finding the perfect amount to cut is a math puzzle that is incredibly hard (NP-hard). It's like trying to find the perfect temperature for a shower where you want it hot enough to be comfortable but not so hot that you get burned, all while the water pressure keeps changing.

The Algorithm: The authors used a smart mathematical tool called Adam Optimization (a type of gradient descent). Think of this as a robot gardener that tries different cutting angles, checks the results, and slowly adjusts its shears until it finds the perfect cut.
The Result: They found that cutting about 30% of the dynamic branches (leaving a pruning factor of 0.7) was the sweet spot.

The Results: A 4x Improvement

When they tested this system with one million transactions (simulating a huge amount of health data):

Old System: The storage cost grew wildly, like a balloon inflating uncontrollably.
New System (DBHMT): The storage cost stayed manageable.
The Win: The new system used four times less storage space than the traditional method.

Security: Did We Cut Too Much?

The authors were very careful to ensure that cutting these branches didn't break the security. They proved mathematically that even with the pruning:

Integrity: You can still prove a record belongs to the tree.
Resistance to Attacks: It is still computationally impossible for a hacker to fake a record or find two different records that look the same (collision).

They created a "safety map" (Pareto front) showing exactly how much storage you save versus how much security you might lose, ensuring the system stays within safe limits (like 128-bit or 256-bit security levels).

Real-World Context (As Stated in the Paper)

The paper specifically mentions that this is designed for IoT-enabled Healthcare 4.0.

The Devices: Wearable fitness trackers, digital thermometers, and pulse oximeters.
The Use Case: Managing Electronic Health Records (EHRs) for both chronic (long-term) and non-chronic (short-term) patients.
The Platform: It is built to work on the IOTA Tangle, a decentralized network that doesn't charge fees for transactions, making it ideal for non-monetary health data.

Summary

The paper presents a new way to organize digital health records that acts like a smart, self-cleaning filing cabinet. By automatically "pruning" (deleting) short-term records while keeping long-term ones, and using a smart algorithm to find the perfect balance, they created a system that is 4 times more efficient at storing data without compromising the security needed to protect patient privacy.

Technical Summary: DBHMT – Dynamic Balancing Hybrid Merkle Tree for IoT-Enabled Healthcare 4.0

1. Problem Statement

The emergence of sixth-generation (6G) technology and the evolution of Healthcare 4.0 have intensified the demand for secure, transparent, and efficient management of Electronic Health Records (EHRs) within resource-constrained Internet of Things (IoT) environments. While decentralized architectures like the IOTA Tangle offer necessary security and transparency, they face significant challenges regarding storage efficiency.

In traditional Healthcare 4.0 scenarios, patient data exhibits dynamic characteristics: some records (e.g., short-term illnesses, injuries) have short lifespans, while others (e.g., chronic conditions) are lifelong. Conventional Merkle tree implementations, which grow super-exponentially with transaction volume, fail to accommodate these dynamics efficiently. As the tree expands to include all historical data, storage costs become prohibitive for IoT devices. Furthermore, existing storage optimization techniques for traditional blockchains (e.g., removing outdated blocks) have not been adequately explored or adapted for the IOTA Tangle's Directed Acyclic Graph (DAG) structure, particularly when balancing the trade-off between storage reduction and cryptographic security guarantees (integrity, pre-image resistance, collision resistance).

2. Methodology

The authors propose a novel data structure called the Dynamic Balancing Hybrid Merkle Tree (DBHMT) designed specifically for the IOTA Tangle. The methodology involves three core components: structural design, mathematical modeling, and optimization.

2.1 Structural Design: The Hybrid Tree

The DBHMT is a vertically partitioned hybrid structure combining an $n$ -ary tree and a binary tree:

Static Side (Right): Represents patients with long-term or lifelong conditions. This side remains static and is not pruned, preserving long-term historical integrity.
Dynamic Side (Left): Represents patients with short-term or transient conditions. This side is highly dynamic, where blocks expire quickly.
Pruning Mechanism: To control the "growing tail" on the dynamic side, the authors introduce a pruning mechanism. Blocks on the left side are removed once they expire, controlled by a pruning factor ( $\alpha$ or $p$ ). This prevents the super-exponential growth typical of standard Merkle trees in high-transaction healthcare scenarios.

2.2 Mathematical Modeling and Security Analysis

The paper derives closed-form analytic expressions for the DBHMT's structural and security properties:

Height and Storage: The effective height and total node count are calculated as functions of the total nodes ( $N$ ), tree levels ( $m, n$ ), and the pruning factor. The storage cost is shown to be a function of the retained dynamic nodes.
Security Constraints: The authors analyze how pruning impacts four critical security properties:
1. Integrity: Resistance to forgery of inclusion proofs.
2. Pre-image Resistance: Difficulty in finding inputs from hash outputs.
3. Second Pre-image Resistance: Difficulty in finding alternative inputs for a given hash.
4. Collision Resistance: Difficulty in finding two inputs with the same hash.
  The analysis establishes that while pruning reduces the attack surface for some properties (e.g., second pre-image), it can degrade others (e.g., pre-image resistance) depending on the retention ratio.

2.3 Optimization Algorithm

Finding the optimal pruning factor is formulated as a non-convex, NP-hard, multi-objective optimization problem. The objective is to minimize storage cost while satisfying security constraints (e.g., maintaining a 128-bit or 256-bit security level).

Objective Function: A composite function balancing height imbalance, total height, storage cost, and security penalties.
Solver: Due to the non-convex nature of the problem, the authors employ a Gradient Descent approach with Adam optimization. This method utilizes first and second moment estimates to accelerate convergence and escape local minima, determining the optimal pruning factor ( $p^*$ ) that satisfies the security constraints while minimizing storage.

3. Key Contributions

The paper outlines four primary contributions:

Novel Architecture: Proposal of the DBHMT, a hybrid Merkle tree tailored for the IOTA Tangle that accommodates the dynamic nature of healthcare data through vertical partitioning (static vs. dynamic).
Dynamic Pruning Concept: Introduction of a dynamic balancing mechanism where the pruning factor is adjustable based on the tree's use case, enabling constant, sub-linear, or linear storage optimization.
Theoretical Framework: Derivation of closed-loop analytic expressions for tree height, storage cost, proof complexity, and security bounds under pruning. This includes a rigorous game-based security proof under the random oracle model.
Optimization Solution: Development of an iterative algorithm using Adam optimization to solve the NP-hard problem of finding the optimal pruning factor, effectively navigating the trade-off between storage efficiency and security.

4. Results and Evaluation

The authors implemented the DBHMT on the IOTA Tangle platform and conducted a proof-of-concept evaluation involving one million transactions.

Storage Efficiency: Experimental results demonstrate that DBHMT achieves a fourfold (4x) reduction in storage cost compared to conventional Merkle trees. This efficiency gap widens as the number of transactions increases, particularly in scenarios with high volumes of short-lived data (e.g., pandemic or flu seasons).
Security Validation: Theoretical analysis and game-based proofs confirm that with an optimal pruning factor (identified around $p \approx 0.7$ in the study), the DBHMT maintains cryptographic strength. The security properties (integrity, pre-image, second pre-image, and collision resistance) remain within acceptable margins (e.g., negligible adversarial advantage for $2^{40}$ queries) even with aggressive pruning.
Convergence: The Adam-based optimization algorithm successfully converged to the optimal pruning factor, demonstrating the feasibility of solving the non-convex storage-security trade-off in dynamic environments.

5. Significance and Claims

The paper claims that DBHMT provides a practical solution for the storage bottleneck in IoT-enabled Healthcare 4.0 without compromising the fundamental security guarantees required for decentralized health records.

Scalability: By enabling storage costs to scale linearly or sub-linearly rather than super-exponentially, DBHMT makes the IOTA Tangle viable for high-volume, resource-constrained IoT healthcare applications.
Flexibility: The framework allows for a tunable trade-off, where practitioners can select a pruning factor that meets specific regulatory or operational security thresholds (e.g., HIPAA compliance) while maximizing storage efficiency.
Broader Impact: The authors suggest that this approach could benefit not only the open-source IOTA community but also traditional blockchain networks (private and consortium) seeking to optimize storage for proprietary applications.

The paper concludes that while the proposed solution significantly reduces storage overhead, it is specifically designed for scenarios where data has a finite lifespan. It acknowledges that for applications requiring permanent, unalterable historical retention, such pruning strategies would not be applicable. Future work is proposed to explore advanced optimization techniques (e.g., Newton's method) and real-world deployment on IOTA 2.0 networks.