Blockchain security failures rarely stem from broken cryptography. They more often arise from incentive incompatibilities, flawed protocol assumptions, and implementation weaknesses that cross the boundaries between architectural layers. Existing research predominantly analyzes consensus, network, and application layers in isolation. This approach often fails to capture vulnerabilities that propagate across layer boundaries. The presented work addresses this gap through a systematic cross-layer security analysis grounded in a Security Reference Architecture for blockchains. The methodology combines game-theoretic modeling; large-scale discrete-event and stochastic simulation; protocol design and prototyping; a systematic survey of 36 cryptocurrency wallets; and zero-knowledge succinct non-interactive arguments of knowledge (zk-SNARKs). At the consensus layer, random transaction selection in Directed Acyclic Graph-based Proof-of-Work protocols is formally proven not to constitute a Nash equilibrium. Simulations with more than 8000 network nodes confirm that discovered incentive vulnerability persists at realistic network scale. A multi-attacker selfish mining evaluation reveals that the profitability threshold in Nakamoto consensus drops from 33% under a single attacker to approximately 14% to 17% per attacker with five simultaneous adversaries, and Strongchain consistently emerges as the most resilient protocol across all tested configurations. At the consensus-network boundary, a Proof-of-Stake protocol with integrated onion routing achieves proposer anonymization at approximately 110 transactions per second. At the application layer, a formal authentication taxonomy reveals that most deployed cryptocurrency wallets provide only single-factor blockchain-verified authorization despite apparent multi-step interfaces. The zk-SNARK-based synchronization architecture reduces mobile client storage from 71MB to 3.5MB and replaces trust in third-party synchronization servers with on-chain cryptographic verification. These contributions collectively demonstrate that blockchain security is an inherently cross-layer property. This research and contributions provide analytical tools, empirical benchmarks, protocol mechanisms, and cryptographic designs for building more resilient decentralized systems.