The Cascadeopt Protocol Sample Clauses

The Cascadeopt Protocol. The connection and difference between the Cascadeopt protocol and our protocol are shown below. ≥ • — The Cascadeopt protocol functions efficiently only when p < 0.20. When p 0.20, the number of passes needed to remove most errors in such a recon- ciliation protocol is larger than 2. Our protocol is more general since it applies to p < 0.5. More precisely, there are t 1 passes in our protocol to make Bob’s bit error probability less than 0.20. • The first two passes of the Cascadeopt protocol correspond to Pass t and Pass t + 1 of our protocol. Our protocol shares with Cascadeopt almost the same principle for determining the optimal block lengths for the two passes. The principle is to minimize the amount of information leaked during the two passes, based on the fact that almost all errors are removed in these two passes. However, the optimal block length wt+1 for Pass t +1 in our protocol is chosen dynamically with the current average bit error rate. It increases during Pass t + 1 rather than staying constant like w2 does in the Cascadeopt protocol. • t Our protocol needs more memory and computational overload to estimate the optimal weights in Pass t and Pass t + 1. The main cost is Algorithm 2.5.2, which has a computational overload of order n′2p(t—1)/2 and memory overload of order n′ to get wopt. Recall that the bit error rates before Pass t can be got from Algorithm 2.5.1, the bit error rate after Pass t can be computed by Equation (2.36) and (2.37), and that after Pass t by Equation (2.38). Since all the bit error rates are got with almost no computational and memory t+1 overload, the dynamic optimal weight wopt during Pass t + 1 can be got from Equation (2.28) almost for free. • The Cascadeopt protocol does not discard any bit. In our protocol, bits (one bit from each block) are discarded with passes going on to compensate the syndrome (parity) information leaked to Eve. So, the working strings are shrinking from pass to pass. Discarding bits also helps error correction because errors may also be discarded. • Simulation results show that our protocol comes up with a better performance (a higher information rate) than the Cascadeopt protocol.