RFID Guardian ACL Testing Effiency

From RFID Guardian

In addition to the software regression test, a special effort was made to test the software e±ciency. In theory, the jamming signals from RFID Guardian can arrive at the readers together with the RFID tags simultaneously, which means that software implementation has to be very e±cient. And whereas the RFID tags just need to do one job (reply), RFID Guardian needs to do three things.

1. Read the Access Control List

2. Read the Tag file

3. Decide whether or not to send a jamming signal

The software achieves high e±ciency by preprocessing, which means that before any queries arrive the software calculates rules for each tags that should be protected. So when the queries arrive, RFID Guardian can use the result of precomputation to decide which tags should protect, thus saving time.

The effect of preprocessing is similar to that of caching, namely that in both cases, a result is stored in memory to save time for future queries. Below is the result of a simulation test to illustrate the bene¯t. In this test a reader broadcasts the same queries 10 times, and the RFID Guardian replies with the same ACL rules and tag sets. The only di®erence is that ¯rst time the software enables preprocessing, but in the second time it does not. The total query time length was measured, with an increasing number of ACL rules to see how behavior changes when the work load becomes heavier. The time is measured in seconds.

Number of Rules	Preprocess	Non-preprocess
1	0.000076	0.000026
2	0.000123	0.000069
3	0.000206	0.000147
4	0.000478	0.000245
5	0.000431	0.000384
6	0.000605	0.000659
7	0.000943	0.000792
8	0.001057	0.000899
9	0.001574	0.000839
10	0.002044	0.002400
20	0.006804	0.006399
30	0.038175	0.021965
40	0.049829	0.041872
50	0.079150	0.083155
60	0.160024	0.128865
70	0.224686	0.225635
80	0.317000	0.352332
90	0.397201	0.465492
100	0.647474	0.924550

When these results are visualized (See Figure 4.1) it becomes apparent that the bene¯t of preprocessing becomes more important as the number of ACL rules grows. If there are 100 rules, preprocessing can save time to as much as 30%!