Internet traffic is increasingly becoming multimedia-centric. Its growth is driven by the fast-growing mobile user base that is more interested in the content rather than its origin. These trends have motivated proposals for a new Internet networking paradigm-information-centric networking (ICN). This paradigm requires unique names for packets to leverage pervasive in-network caching, name-based routing, and named-data provenance. However named-data routing makes user censorship easy. Hence an anti-censorship mechanism is imperative to help users mask their named queries to prevent censorship and identification. However, this masking mechanism should not adversely affect request rates. In this paper, we propose such an anti-censorship framework, which is lightweight and specifically targets low compute power mobile devices. We analyze our framework's information-theoretic secrecy and present perfect secrecy thresholds under different scenarios. We also analyze its breakabil-ity and computational security. Experimental results prove the framework's effectiveness: for requests it adds between 1.3-1.8 times in latency overhead over baseline ICN; significantly lesser than the overhead of the state of the art Tor (up to 38 times over TCP).