1. A Dynamic Indirect IP Lookup based on Prefix Relationships
Author: Yu-Chen Kuo, Chih-Cheng Li
Publisher: IEEE APCC 2012
Presenter: Sih-An Pan
Date: 2014/1/8
一樣有個16bit的segment table 根據Prefix關係去紀錄endpoint endpoint 用bitmap做壓縮
Small Forwarding Table 不能update
many indirect lookup schemes [8~10] that had the advantages of using less memory and require less memory access

2. INTRODUCTION
We adopt a compressed bitmap data structure for IP lookup and develop a prefix parent relationship data structure for dynamically update prefixes.
Most of previously proposed indirect schemes were suitable only for static routing tables, but were not suitable for dynamic routing tables where frequent updates are required.

3. THE PROPOSED SCHEME 分3種情況Maximum prefix length
Maximum prefix length less than or equal to 16 in the segment 對這個segment而言 那個prefix就一定是LPM
Compression等講完下一頁再回頭講
For example, if the maximum prefix length is 20 in the segment, the CNH equals 4 and code word table has only
one entry.
最大length 最小範圍

4. THE PROPOSED SCHEME This will allow IP lookup to be performed more
efficiency without comparing data in layers.
Fi>Fj
Pi Pj disjoin

5. THE PROPOSED SCHEME

6. THE PROPOSED SCHEME

7. THE PROPOSED SCHEME

8. Search algorithm

9. Delete in our scheme
Segment and offset are derived from deleted prefix, and corresponding EL of offset can be determined. The segment(s) due for processing can also be determined from the prefix length.
Find the deleted PN and its child PN. Let child PN’s parent PN link to deleted PN’s parent PN. If the EL links to deleted PN, replace the link in ELs with deleted PN’s parent PN.
If continuous EL links are present, duplicate entries are deleted and retain one entry.
Delete the PN.
刪掉的prefix可能prefix length是15 那會存在2個Segment 都要處理 CWT可能會變

10. PERFORMANCE ANALYSIS

11. PERFORMANCE ANALYSIS

12. PERFORMANCE ANALYSIS