A Peer-to-Peer Filter-Based Algorithm for Internal Clock Synchronization in Presence of Corrupted Processes

Abstract

This paper proposes an internal clock synchronization algorithm for very large number of processes that is able to (i) self-synchronize their local clocks without any central control and (ii) resist to attacks of an adversary whose aim is to put out-of-synchronization as many correct processes as possible. To cope with scale the algorithm utilizes the gossip-based paradigm where each process has a limited view of the system, while to resist to attacks the algorithm employs a filtering mechanism based on the notion of α-trimmed mean to filter out out-of-range clock values. The algorithm shows nice convergence in presence of networks errors and in absence of the adversary. When the adversary takes control of some of the processes in the system, we define two goals for the adversary, actually two predicates, to measure the strength of the attack. The first one captures the percentage of time in which at least one correct is out of synchronization and the second one when all correct processes are out of synchronization. The paper presents an extensive simulation study showing under which conditions (in terms of number of corrupted processes and size of local views) these two goals can be achieved by the adversary. Interestingly, these results can be exploited by applications that can tolerate either a certain time in which some correct process is non-synchronized or a certain percentage of correct processes that is non-synchronized.

Hardening Cassandra Against Byzantine Failures

“…For time synchronization, Cassandra requires the administrator to provide an external solution such as NTP. In our work, we follow this guideline using the latest version of NTP that can tolerate Byzantine faults when ensuring the usage of SSL and authentication measures [6,45]. We configure this service so that all servers could use it as is and clients would be able only to query it, without affecting the time.…”

Time synchronization in the DHT-based P2P network Kad for real-time automation scenarios

“…In [11], a system is presented, which uses a gossip-based approach. The focus is on assessing the impact of corrupted processes on the effectiveness of clock synchronization.…”

“…Several parameters have to be determined to achieve a high probability of information exchange. Using a gossip-based or epidemic approach like in [11] and [12] is always a trade-off between scalability and reliability [13]. Therefore, we renounce a gossip-based approach as it is probabilistic and not suited for hard real-time scenarios requiring deterministic behavior of the network.…”

“…Table II is competitive to existing central solutions like NTP with an accuracy of several milliseconds whereby a maximum time error T M axError of 1 ms is assumed. Furthermore, our approach is not gossip-based like the solutions in [11] and [12] and therefore it is suitable to be applied in a hard real-time scenario. The results have been proven by simulations and confirm the values for T SynComp and T ReSyn , which have been previously calculated.…”