### Table 1. Verification results

2004

"... In PAGE 11: ... Note that spin carries out a step- by-step execution, allowing the inspection of code and variables in every step. The size of the state space produced by this model is shown in Table1 (row safety in concrete model). Note that the state space increases exponentially with the number of floors in the lift.... In PAGE 16: ... 15, with this transformation the error NoMove is not present in either the abstract model or, using Theorem 1, the concrete model. The benefits of using the abstract formula to discard this error are sum- marized in Table1 and Fig. 19.... In PAGE 16: ... Now the verification result for the abstract formula is valid , which ensures (using Theorem 1) that the con- crete model satisfies the property. Again, the benefits of verifying with this method are shown in Table1 and Fig. 19.... ..."

Cited by 2

### Table 1. Standard specifications for turbo decoding

"... In PAGE 11: ... 11 Yeo, Nikolic, Anantharam 5/30/2003 5. Platforms for Iterative Decoding Table1 lists some recent communication standards specifying the use of iterative decoding for forward error correction. Iterative codes are also currently being considered for high performance storage applications that require about 1Gbps throughput as well as 10Gbps optical communications.... ..."

### TABLE IX MODIFIED RATE 1/2 STRUCTURE FOR LOWER ERROR FLOOR. Variable Constraint

### Table 5. Single-floor median localization error (meters).

2005

"... In PAGE 12: ... This is strong evidence that extending fingerprints to include signal strength information from channels other than the 6-strongest cells, even when the identity of the transmitter cannot be determined, can dramatically increase localization accuracy. Within-Floor Localization Error Table5 summarizes the localization errors within specific floors for the 5 algorithms introduced in Section 3.4 for the three indoor environments.... In PAGE 12: ... All calculations assume a training set restricted to include only points that are on the same floor as the point whose position is being determined. Table5 also presents results for random, an algorithm that determines lo- calization by picking an arbitrary position in the building. Therefore, random provides a lower bound on the performance of localization systems for a given floor and building.... ..."

Cited by 16

### Table 5. Single-floor median localization error (meters).

2005

"... In PAGE 12: ... This is strong evidence that extending fingerprints to include signal strength information from channels other than the 6-strongest cells, even when the identity of the transmitter cannot be determined, can dramatically increase localization accuracy. Within-Floor Localization Error Table5 summarizes the localization errors within specific floors for the 5 algorithms introduced in Section 3.4 for the three indoor environments.... In PAGE 12: ... All calculations assume a training set restricted to include only points that are on the same floor as the point whose position is being determined. Table5 also presents results for random, an algorithm that determines lo- calization by picking an arbitrary position in the building. Therefore, random provides a lower bound on the performance of localization systems for a given floor and building.... ..."

Cited by 16

### Table 5. Single-floor median localization error (meters).

"... In PAGE 12: ... This is strong evidence that extending fingerprints to include signal strength information from channels other than the 6-strongest cells, even when the identity of the transmitter cannot be determined, can dramatically increase localization accuracy. Within-Floor Localization Error Table5 summarizes the localization errors within specific floors for the 5 algorithms introduced in Section 3.4 for the three indoor environments.... In PAGE 12: ... All calculations assume a training set restricted to include only points that are on the same floor as the point whose position is being determined. Table5 also presents results for random, an algorithm that determines lo- calization by picking an arbitrary position in the building. Therefore, random provides a lower bound on the performance of localization systems for a given floor and building.... ..."

### Table 2: Average EER Verification performance on the YOHO database.

"... In PAGE 4: ... Also note that none of the methods use a pooled background model which has been shown to perform better than cohort normalization [6]. If we compare the parameter-usage performance of our sys- tem, then the systems reported in Table2 use BJBEBKBC, BHBEBCBC, BGBIBKBCBC, and BDBDBEBGBK parameters, respectively. We estimate the pa- rameter usage of [2] (for example) for the BF-mixture case with BEBC monophone models to be about BDBGBEBEBC parameters.... In PAGE 4: ... We estimate the pa- rameter usage of [2] (for example) for the BF-mixture case with BEBC monophone models to be about BDBGBEBEBC parameters. Thus, if we consider the second system in Table2 , we have a lower error rate and use BIBFB1 less parameters. We also addressed the issue of bias from training against known impostors.... In PAGE 4: ... Veri- fication was performed by using the second BIBL as impostors for the first BIBL and vice versa. The resulting average EER increase for the BDBE MFCC, BDBE A1-MFCC, degree BE system in Table2 was from BCBMBEBLB1 to BCBMBFBHB1; this small increase shows that the system generalizes well to unknown impostors. For a computationally scalable identification system, we want the computation growth with the number of speakers, CRC6 D7D4CZ , to have as small of CR as possible.... ..."

### Table 6.2: Turbo decoder design characteristics before and after applying the CRC-HDD dynamic-iterative technique

2005

### Table 3. XOR values for calculating the CRC.

"... In PAGE 6: ...This eight bit code is a CRC code. Briefly, it is calculated by XOR-ing the values in Table3 for every bit in the datapacket that is 1. This code can also be represented with CRC polynomial x8 + x6 + x5 + x3 + x +1.... ..."