FibTree: Merkle Tree with Fibonacci Branching for Delay-Tolerant Networks — Theoretical Analysis, Experimental Validation, Optimizations, and End-to-End Encryption
Fabio Pimenta Pinto, Flávia Christini Lima Pinto · Zenodo (CERN European Organization for Nuclear Research) · 2026
This work proposes FibTree, a hierarchical data structure that combines Merkle trees with Fibonacci-based branching, designed for efficient data distribution in Delay-Tolerant Networks (DTN). FibTree organizes messages in a hierarchy of levels where each node groups F(n) children, applying Reed-Solomon coding independently at each level. Version 3.0 extends the previous work with a fourth contribution: end-to-end encryption (E2EE) integrated at the application layer, allowing messages to be decrypted only by their intended recipient, even when intermediate nodes forward the fragments. Experimental results confirm the analytical predictions: FibTree achieves 98.9% delivery in channels with 50% loss, versus 14.5% for flat Reed-Solomon, a gain of 84.4 percentage points. FibTree's average latency (136 to 187 ms) is approximately 6 times lower than that of RS(10,15) (985 to 1126 ms). The combination of three optimization techniques reduces overhead from 776.7% to values between -33.8% (bandwidth savings) and 128.1%, while fully preserving delivery rate. The E2EE layer, implemented using ECDH X25519, HKDF, and AES-256-GCM, was validated through three scenarios proving that only the intended recipient can decrypt the message. This layer was proposed by co-author Flávia Christini de Lima Pinto, whose question about privacy in peer-to-peer networks redirected the entire research effort. Seven concrete industrial bottlenecks were identified, with emphasis on AI model distribution, OTA updates in vehicles, and space communication. The reference implementation is publicly available in Python and Kotlin.