Orthogonal arrays in statistics and computer science
Most prominent among the applications of error-correcting codes is the transmission of messages via noisy channels. From this point of view the minimum distance is the most important parameter. In the linear case the duality theorem leads to a different approach to codes and to a different type of application. Any subset of random variables will then be statistically independent if and only if the corresponding columns of the generator matrix are linearly independent. The most prominent statistical application of orthogonal arrays is in the design of experiments. Linear shift register sequences, the most popular and frequently used family of pseudorandom sequences, are the object. This chapter discusses the use of distance and strength in cryptography. It describes an application of OA of strength 2 in the testing of chips.