This paper focuses on the security and reliability issues of communication protocols for distributed, large scale and dense networks. In particular, the resilience of routing protocols is investigated against malicious insiders willing to disrupt network communications. Previously, the simulation results showed that introducing randomness in routing protocols increases uncertainty for an adversary, making the protocols unpredictable. When combined with data replication, it permits route diversification between a source and a destination, thus enhancing the resilience. In this paper, we propose a theoretical framework to quantify analytically the performance of random protocols against attacks based on biased random walks on a torus lattice. The objective is to evaluate analytically the influence of bias and data replication introduced to random walks. The bias allows to decrease the route length by directing random walks toward the destination, thus reducing the probability of a data packet to meet a malicious insider along the route; however, it decreases also the degree of randomness (entropy). When random protocols are combined with data replication, the reliability is improved thanks to route diversity despite an additional overhead in terms of energy consumption. The main goal is to provide a good tradeoff between shortest path and route diversity for a reasonable cost.