In classical cryptography, threshold secret sharing schemes—such as Shamir’s $\left(t, n\right)$ protocol—operate under a strict assumption: the architecture of the group is frozen at the moment of initialization. You choose a fixed number of total participants ($n$) and a fixed threshold ($t$) required to reconstruct the secret.
However, modern decentralized networks, blockchains, and ad-hoc distributed systems do not live in a static world. Nodes frequently churn, new participants enroll sequentially, and the financial or operational value of a protected asset may fluctuate over time, necessitating a tighter or looser security threshold.
As noted by Liaojun Pang et al. in An Identity (ID)-Based Threshold Multi-Secret Sharing Scheme and Yulian Shang et al. in Application of the Central Encryption System in the Dynamic Secret Sharing Scheme (II), forcing a full system reset or redistributing private shares every time the group configuration changes introduces massive communication overhead and creates a single point of failure during the redistribution phase.
To solve this, a robust body of research has converged on a fluid paradigm: Dynamic and Evolving Secret Sharing. This approach allows systems to infinitely enroll new participants, alter reconstruction thresholds on the fly, and update group boundaries without ever redistributing private keys or resetting the underlying secret.
Evolving Access Structures for Infinite Growth
Imagine a network where participants join sequentially over an infinite time horizon. In a standard setup, the dealer must know the final population size $n$ in advance to construct the polynomial. If the dealer initializes the system with $n=100$, the 101st participant cannot be accommodated.
To bypass this limitation, Ilan Komargodski et al. formalized the concept of evolving access structures in How to Share a Secret, Infinitely. The core intuition relies on grouping participants into chronological generations. Instead of using a single static polynomial, the system employs recursive composition. When a new generation of users joins, their shares are linked to the previous generation's structure using algebraic combinations or prefix codes over polynomial rings, a mechanism further refined by Hongru Cao et al. in A Construction of Evolving k-Threshold Secret Sharing Scheme over A Polynomial Ring.
Generation 1 (Users 1-3) ---> Standard (t, 3) Scheme
│
Generation 2 (Users 4-7) ---> Linked via Recursive Composition
│
Generation 3 (Users 8-12) ---> Prefix Codes over Polynomial Ring
Early iterations of evolving secret sharing suffered from a major drawback: to maintain information-theoretic ("perfect") secrecy, the size of the shares given to newer participants had to grow exponentially or quartically relative to the pool size.
The breakthrough that made infinite enrollment practical was the shift from information-theoretic security to computational security. In Evolving Secret Sharing Made Short and Evolving secret sharing revisited: computational security and succinctness, Danilo Francati et al. demonstrated that by leveraging cryptographic primitives like Projective Pseudorandom Generators (pPRGs), share sizes could be kept succinct. Instead of storing the entire history of the evolving access tree, a new participant's share grows merely linearly with the threshold complexity. The mathematical intuition is elegant: rather than distributing raw, perfectly insulated algebraic pieces, the dealer distributes short cryptographic seeds that can deterministically expand into authorized evaluation points, bypassing classical lower bounds on share size.
Modifying Thresholds on the Fly
Sometimes the number of participants remains stable, but the environment demands a change in the threshold $t$. For instance, if an adversary becomes more powerful, an organization might want to increase the threshold from $t=3$ to $t=5$ to prevent collusion. Conversely, during an emergency or network partition, they might need to decrease it to maintain availability.
In Lattice-Based Threshold-Changeability for Standard Shamir Secret-Sharing Schemes, Ron Steinfeld et al. introduced a communication-free protocol for threshold modification. Instead of relying on an active central dealer to compute a new polynomial, existing shareholders perform local linear transformations on their shares. By multiplying their shares by specific, locally derived lattice coefficients, they effectively increase the degree of the implicit polynomial across the network without interacting with one another.
For arbitrary adjustments—where the threshold needs the flexibility to move both up and down—Lifeng Yuan et al. proposed a framework in Novel Threshold Changeable Secret Sharing Schemes Based on Polynomial Interpolation utilizing two-variable one-way functions. By evaluating a bivariate polynomial $F(x, y)$ where one variable handles the participant's identity and the other handles the active threshold state, the system can smoothly transition between security levels.
To minimize the data overhead inherent in these transitions, Rawad Bitar et al. designed Staircase Codes in Staircase Codes for Secret Sharing With Optimal Communication and Read Overheads. Staircase codes structure the shared data matrix such that users only need to read and download the absolute minimum fractional pieces of information required for a given threshold state, optimizing both storage and communication costs universally across all possible threshold configurations.
This dynamic thresholding is highly applicable to real-world systems. For example:
- Decentralized Committees: In CHURP: Dynamic-Committee Proactive Secret Sharing, Deepak Maram et al. utilized asymmetric bivariate polynomials to temporarily alter thresholds during blockchain committee handoffs, protecting the state while nodes churn.
- Homomorphic Encryption: In Arbitrary-Threshold Fully Homomorphic Encryption with Lower Complexity, Songze Li et al. leveraged dynamic subset thresholding to allow secure multi-party decryption to proceed with any subset meeting a fluid threshold, ensuring system robustness if nodes drop offline mid-protocol.
Decoupling Identity via Public Bulletins and Chains
How do we enroll or remove a specific participant without triggering an "all-or-nothing" redistribution problem? The answer lies in decoupling a user's private cryptographic identity from the mathematical evaluation of the secret.
In A Dynamic Secret Sharing Scheme Based on Factorization, Yufei Zhang introduced an architecture centered around a public bulletin board. Instead of giving users raw polynomial coordinates that map directly to their identities, the system stores transformed polynomial evaluations on a public ledger.
$$\text{Private Share } (s_i) \ \ + \ \ \text{Public Bulletin Entry } (B_i) \ \ \longrightarrow \ \ \text{Valid Polynomial Point } (x_i, f(x_i))$$
When a new member joins or an old member is revoked, the system does not issue new private keys. Instead, it updates only the relevant public entries on the bulletin board. A user's private share remains static; its algebraic meaning is dynamically shifted by modifying the public bulletin value. This eliminates the administrative overhead of private key redistribution.
An alternative mechanism for lightweight enrollment is the chained XOR mechanism proposed by Zhihui Wang et al. in An Adaptable (n, n) Secret Image Sharing Mechanism Based on Boolean Operation. Designed for rapid, low-compute environments, this scheme avoids heavy polynomial interpolation entirely. It links participants together via a sequential chain of logical XOR operations. To enroll a new participant, the system simply prepends a new random matrix to the front of the existing chain and updates the sequence tags. The existing shadow shares remain perfectly valid, allowing modular updates in multi-user environments.
Finally, this concept of late-binding authorization was extended to the message level by Hao Xiong et al. in Continual Leakage-Resilient Dynamic Secret Sharing in the Split-State Model. Their scheme allows a sender to specify both the authorized participant set and the reconstruction threshold at the exact moment of message generation, rather than during system setup. By combining this with a split-state model that periodically refreshes shares, the system remains secure even against continuous side-channel leakage.
Similarly, Jingwei Liu et al. demonstrated in MDBV: Monitoring Data Batch Verification for Survivability of Internet of Vehicles that integrating dynamic state parameters directly into cryptographic operations allows entities to enter or exit specific network domains seamlessly, ensuring that security boundaries remain as fluid as the physical or logical groups they protect.
Key Foundational Work
Lattice-Based Threshold-Changeability for Standard Shamir Secret-Sharing Schemes (2004) — contributed 2 concept elements to this direction.
Cryptanalysis and improvement of an ID-based ad-hoc anonymous identification scheme at CT-RSA 05 (2005) — contributed 1 concept elements to this direction.