In 2008, S. Nakamoto presented Bitcoin as a peer-to-peer version of electronic cash -- ^a system allowing direct payments between participants without the need for a financial institution or trusted third party. Thereby, he proposed the first practical solution for the problem of reaching consensus in a dynamic set of potentially anonymous participants without a prior agreement on this set. Bitcoin achieves this advancement at the cost of high computational requirements for Proof-of-Work, leading to vast amounts of electricity being consumed. Recently, new protocols, using Proof-of-Stake as a fundamental principle, have tried to improve upon Nakamoto's solution. These protocols require a trustworthy source of randomness, i.e. publicly-verifiable and bias-resistant randomness, to maintain desirable security guarantees. However, obtaining trustworthy randomness, in a highly decentralized network and under potentially adversarial conditions, is by itself a challenging task. ^Recent academic research as well as projects from the industry try to address this problem by designing random beacon protocols, which produce the required random values in regular intervals. In this thesis, we highlight the design challenges of random beacon protocols as well as provide the first in-depth review and comparison of state-of-the-art protocols. We identify public-verifiable secret sharing (PVSS) as a common building block and develop a new protocol using this cryptographic primitive. Our PVSS-based random beacon protocol greatly improves upon the scalability of existing approaches. Communication complexity is reduced from O(n) to O(n), as our solution only requires a single PVSS instance per round. Our protocol achieves this advancement while ensuring important protocol characteristics such as public-verifiability, bias-resistance and unpredictability. ^Additionally, we present an optimized solution as a protocol extension, which further reduces the interaction between network nodes and achieves a near optimal communication complexity of O(n c). This improvement is accomplished while still retaining liveness, bias-resistance and unpredictability of the random beacon values with very high probability.