China
Africa
Explore chapters and articles related to this topic
Security and Privacy in Vehicular Networks Challenges and Algorithms
Published in Anna Maria Vegni, Dharma P. Agrawal, Cognitive Vehicular Networks, 2018
Yi Gai, Jian Lin, Bhaskar Krishnamachari
Advanced data fusion techniques are required to defend against SSDF attack. Specifically, the data fusion mechanism must discriminate all sensor terminals with their trust or reputation level. Ideally, only the data from reliable sensing devices should be accepted while unreliable devices should be filtered. In [50], a two-level defense against SSDF attack, called weighted sequential probability ratio test (WSPRT), is proposed. The first step is a reputation maintenance step, and the second step is the actual hypothesis test, in which the data collector authenticates sensing reports. The purpose of the first level is to prevent relay attacks or falsified data injected from outside of the CR network. All CRs have an initial reputation value of zero, which increments upon each correct local data report. The hypothesis test step is based on sequential probability ratio test (SPRT), which is a hypothesis test for sequential analysis and supports sampling a variable number of observations. The SPRT takes into account each CR’s reputation value. The WSPRT method has several drawbacks, which include the requirement of a large number of samples and thus long sensing time, the possibility of leading to a deadlock with no decision made, and low performance in a highly dynamic environment. To overcome these drawbacks, the authors in [51] presented two enhanced schemes, named enhanced weighted sequential probability ratio test (EWSPRT) and enhanced weighted sequential zero/one test (EWSZOT). Specifically, EWSPRT adopts a more aggressive weight module while using the same test module as WSPRT. EWSZOT uses the same aggressive weight module as EWSPRT and a new sequential 0/1 test module instead of SPRT. It is shown through simulation that the sampling numbers of EWSPRT and EWSZOT are 40% and 75% lower than WSPRT [50], with comparable errors rates.
Case study on applying sequential analyses in operational testing
Published in Quality Engineering, 2023
Monica Ahrens, Rebecca Medlin, Keyla Pagán-Rivera, John W. Dennis
Alternatively, one could consider conducting the test using a sequential approach. One of the best-known applications of sequential analysis involves testing a hypothesis when the final sample size is not fixed at the start of the analysis; instead, it depends on the information obtained as the data are collected. This procedure underlies the genesis of sequential analysis as formalized by Wald (1945) in his SPRT. The SPRT involves taking observations one at a time; each additional observation is used to decide whether to stop sampling and accept or reject the null hypothesis in question. Wald notes that the SPRT requires, in general, a considerably smaller expected number of observations than the fixed number of observations required by a corresponding non-sequential test, while controlling the type I and type II errors.