Access to information is crucial during conflicts and other critical events such as population uprisings. An increasing number of social interactions happen in the cyberspace, while information exchanges at the infrastructural level (monitoring systems, sensor networks, etc.) are now also based on Internet and wireless links rather than ad hoc, isolated wired networks. However, the nature of the Internet allows powerful hostile actors to block, censor, or redirect communication to and from specific Internet services, through a number of available techniques.Anonymity networks such as Tor provide a way to circumvent traditional strategies for restricting access to online resources, and make communication harder to trace and identify. Tor, in particular, has been successfully used in past crises to evade censorship and Internet blockades (Egypt in 2011, and Iran in 2012). Anonymity networks can provide essential communication tools during conflicts, allowing information exchanges to be concealed from external observers, anonymised, and made resilient to imposed traffic controls and geographical restrictions. However, the design of networks such as Tor makes them vulnerable to large-scale denial of service attacks, as shown by the DDoS targeted at Tor hidden services in March 2015.In this paper, we analyse the structural weaknesses of Tor with regard to denial of service attacks, and propose a number of modifications to the structure of the Tor network aimed at improving its resilience to a large coordinated offensive run by a hostile actor in a conflict scenario. In particular, we introduce novel mechanisms that allow relay information to be propagated in a distributed and peer-to-peer manner. This eliminates the need for directory services, and