The development of blockchain, along with the creation of Bitcoin, opened the road to the first digital currency that is secure against double-spending without the need for a trusted authority or centralized server. The architecture of traditional blockchain structures like the one used in Bitcoin is, however, not very scalable because of its limiting slow global consensus mechanism.
To address this limitation multiple paths have been explored and one of these is represented by Trustchain. Its architecture is based on a pair-based ledger system; where pairs of nodes register their transactions on entangled blocks stored in local, immutable and ordered chain. This technology guarantees high transaction throughput but has a big drawback; it is not possible to actively prevent fraudulent operations like double-spending but only to detect them a posteriori. The goal of this thesis is to research and evaluate mechanisms that can be implemented to reduce the detection time of this fraudulent activity, in a system like Trustchain. The first strategy involve relying on global dissemination, but this is quite inefficient. The second one uses anonymous auditing and localized dissemination to better perform regarding efficiency, at a small expense in terms of transaction duration. The performance improvements are confirmed by simulations. If compared to the existing mechanism, the second proposed solutions achieved double-spending detection in just a fraction of the original time at a cost of a small overhead.