Enabling anytime and anywhere random data access for the mobile sinks is an important solution to the problem in the large scale wireless sensor networks.However,the data discovery in large scale sensor networks faces a non-trivial challenge because of the double blindness between the data and the sinks.The existing schemes are not able to provide the high success rate and consume less energy at the same time.In this paper,the tradeoff between randomness and structure is explored.Based on random line walk scheme,three scalable and efficient data discovery methods are presented to solve the double-blindness.These methods utilize simple geometric property to construct the path of data cache and query and can work well without location.The extensive simulations,using ns-2 and theoretical analysis,have proven the efficient trade off between randomness and structure.