IP-enabled sensors can be globally addressable by any Internet-connected entity, and therefore, their protection presents different challenges than that of traditional sensors, as they are subject to any potential attacker in the Internet. For this reason, specific security protocols must be developed to address the security requirements of IP-enabled sensors. An interesting approach to achieve this aim is the Ladon security protocol, which allows resource-deprived devices to efficiently implement end-to-end authentication, authorisation and key establishment mechanisms. However, in so limited environments such as sensor networks, not only efficient protocols must be defined, but they must also be optimally parameterised. This paper constitutes a step forward in this direction. First, a state transition model of the Ladon protocol is presented to analytically describe its behaviour. Then, this model is used to select the most effective parameterisation of the protocol in terms of message retransmissions and execution of cryptographic operations. The obtained results show that the selected parameterisation allows maximising the probability of a successful secure session establishment, while keeping the overhead introduced by the protocol low. Additionally, the performance comparison carried out shows that Ladon outperforms alternative approaches to achieve the same objective in terms of message transmission and reception operations.