Erasure Code-based Fast Failure Recovery Scheme for Survivable Network
碩士 === 國立中正大學 === 通訊工程研究所 === 99 === When a network failure happens, data going through the failure device is lost. The lost data can be viewed as they are erased on the way to the destination node. Based on this concept, in this thesis, we consider communication paths in a network as erasure channe...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | en_US |
Published: |
2011
|
Online Access: | http://ndltd.ncl.edu.tw/handle/96629654547568144446 |
id |
ndltd-TW-099CCU00650034 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-099CCU006500342015-10-13T20:09:14Z http://ndltd.ncl.edu.tw/handle/96629654547568144446 Erasure Code-based Fast Failure Recovery Scheme for Survivable Network 使用糾刪碼設計具快速故障修復之高存活性網路 Wen-Yu Chang 張雯毓 碩士 國立中正大學 通訊工程研究所 99 When a network failure happens, data going through the failure device is lost. The lost data can be viewed as they are erased on the way to the destination node. Based on this concept, in this thesis, we consider communication paths in a network as erasure channels and apply erasure code for network protection. The benefit of using erasure code for network protection is that no fault management including failure detection, protocol messages exchanges, and traffic rerouting are required in the intermediate nodes. A node that encounters a failure does not need to perform any reaction to it. Those missed information can be recovered based on the remaining data received at the destination nodes. In addition, the network does not need to distinguish transient failure and long lived failure. It avoids network state flipping for handling transient failure. To apply erasure code in network protection against failure, multiple paths between a pair of source node and destination node are needed to provide diversity and bandwidth. More specifically, one needs to jointly determine routing paths, path rates, and coding scheme while to meet network capacity constraint. In this work, we investigate using the most powerful erasure code, Maximum Distance Separable (MDS) codes for network protection. Besides applying basic MDS code in our design, we also consider using XOR code. The benefit of using XOR code is its easiness for encoding and decoding, however, the provisioned paths need to be mutually disjointed. In this thesis, we have formulated the MDS network protection problems using integer linear programming models where the objective functions are to minimize the total bandwidth consumption in the network. To make performance comparison, besides the proposed basic MDS code network protection, we also implement conventional 1+1 path protection that is also a fault management free protection scheme for intermediate nodes. We draw performance comparisons via performing a lot of numerical experiments on three benchmark networks. The experimental results show that MDS based protection outperforms 1+1 path protection in all test cases. In addition, we also compare XOR First MDS protection with non-XOR code protection. The results indicate that using only XOR code for all communication pairs is good enough for almost all cases. Only very small bandwidth penalty is paid for XOR code. Finally, we investigate the proposed schemes for network with dynamic routing. Comparing both schemes, we find that enhanced MDS code protection has lower blocking probability and bandwidth blocking rate than pure XOR code. Steven S. W. Lee 李詩偉 2011 學位論文 ; thesis 42 en_US |
collection |
NDLTD |
language |
en_US |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立中正大學 === 通訊工程研究所 === 99 === When a network failure happens, data going through the failure device is lost. The lost data can be viewed as they are erased on the way to the destination node. Based on this concept, in this thesis, we consider communication paths in a network as erasure channels and apply erasure code for network protection. The benefit of using erasure code for network protection is that no fault management including failure detection, protocol messages exchanges, and traffic rerouting are required in the intermediate nodes. A node that encounters a failure does not need to perform any reaction to it. Those missed information can be recovered based on the remaining data received at the destination nodes. In addition, the network does not need to distinguish transient failure and long lived failure. It avoids network state flipping for handling transient failure.
To apply erasure code in network protection against failure, multiple paths between a pair of source node and destination node are needed to provide diversity and bandwidth. More specifically, one needs to jointly determine routing paths, path rates, and coding scheme while to meet network capacity constraint. In this work, we investigate using the most powerful erasure code, Maximum Distance Separable (MDS) codes for network protection. Besides applying basic MDS code in our design, we also consider using XOR code. The benefit of using XOR code is its easiness for encoding and decoding, however, the provisioned paths need to be mutually disjointed.
In this thesis, we have formulated the MDS network protection problems using integer linear programming models where the objective functions are to minimize the total bandwidth consumption in the network. To make performance comparison, besides the proposed basic MDS code network protection, we also implement conventional 1+1 path protection that is also a fault management free protection scheme for intermediate nodes. We draw performance comparisons via performing a lot of numerical experiments on three benchmark networks. The experimental results show that MDS based protection outperforms 1+1 path protection in all test cases. In addition, we also compare XOR First MDS protection with non-XOR code protection. The results indicate that using only XOR code for all communication pairs is good enough for almost all cases. Only very small bandwidth penalty is paid for XOR code.
Finally, we investigate the proposed schemes for network with dynamic routing. Comparing both schemes, we find that enhanced MDS code protection has lower blocking probability and bandwidth blocking rate than pure XOR code.
|
author2 |
Steven S. W. Lee |
author_facet |
Steven S. W. Lee Wen-Yu Chang 張雯毓 |
author |
Wen-Yu Chang 張雯毓 |
spellingShingle |
Wen-Yu Chang 張雯毓 Erasure Code-based Fast Failure Recovery Scheme for Survivable Network |
author_sort |
Wen-Yu Chang |
title |
Erasure Code-based Fast Failure Recovery Scheme for Survivable Network |
title_short |
Erasure Code-based Fast Failure Recovery Scheme for Survivable Network |
title_full |
Erasure Code-based Fast Failure Recovery Scheme for Survivable Network |
title_fullStr |
Erasure Code-based Fast Failure Recovery Scheme for Survivable Network |
title_full_unstemmed |
Erasure Code-based Fast Failure Recovery Scheme for Survivable Network |
title_sort |
erasure code-based fast failure recovery scheme for survivable network |
publishDate |
2011 |
url |
http://ndltd.ncl.edu.tw/handle/96629654547568144446 |
work_keys_str_mv |
AT wenyuchang erasurecodebasedfastfailurerecoveryschemeforsurvivablenetwork AT zhāngwényù erasurecodebasedfastfailurerecoveryschemeforsurvivablenetwork AT wenyuchang shǐyòngjiūshānmǎshèjìjùkuàisùgùzhàngxiūfùzhīgāocúnhuóxìngwǎnglù AT zhāngwényù shǐyòngjiūshānmǎshèjìjùkuàisùgùzhàngxiūfùzhīgāocúnhuóxìngwǎnglù |
_version_ |
1718044281378177024 |