Performance Analysis for Finite Length LT Codes via Classical Probability Evaluation

Suo, Longlong; Zhang, Gengxin; Lv, Jing; Xiang, Tian

doi:10.1109/lcomm.2017.2700303

Cited by 11 publications

(4 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The necessary number of encoded packets to recuperate initial information, with K input symbols used in LT encoding, is 22,32…”

Section: Classification Methods and Decisionmentioning

confidence: 99%

“…The average number of packets m opt to decode the message and recuperate initial information is the solution of the following optimization problem:

\underset{m}{argmax} true \sum_{i = 1}^{m} P false(C p_{j} false| p_{i} false), j = 1, 2,

s.t. c_{i} = 0 .

The necessary number of encoded packets to recuperate initial information, with K input symbols used in LT encoding, is

N = K + scriptO false(\sqrt{K} \log^{2} K false) .

…”

Section: System Model and Problem Formulationmentioning

confidence: 99%

See 1 more Smart Citation

Exploiting classification for fountain data estimation in wireless sensor networks

Belabed

Bouallègue

2020

Int J Communication

View full text Add to dashboard Cite

Summary In order to correct and avoid channel error, fountain codes were the best solution by limiting feedback channels and reducing energy consumption. Multi‐hops transmission is the principal limitation of the deployment and the use of these codes. Indeed, relayed transmission conducts to the generation of useless data, named overflow leading to a waste of energy, the most critical issue, and the big challenge in WSN. In this paper, based on a clustered architecture and estimation, we consider a distributed estimation scheme composing of sensor members and the cluster head. In order to reduce the number of a useless encoded packet generated as well as the impact of the overflow, we determine the optimal minimal number of encoded packets needed for data decoding. Sensor observations are encoded using fountain codes, and then messages are collected at the cluster head where a final estimation is provided within learning method. Then messages are collected at the cluster head where a final estimation is provided with a classification based on Bayes rule. The main goal of this paper is to determine the number of encoded packets by exploiting the classification model for fountain data estimation to minimize the overflow and extend the network lifetime.

show abstract

“…The necessary number of encoded packets to recuperate initial information, with K input symbols used in LT encoding, is 22,32…”

Section: Classification Methods and Decisionmentioning

confidence: 99%

“…The average number of packets m opt to decode the message and recuperate initial information is the solution of the following optimization problem:

\underset{m}{argmax} true \sum_{i = 1}^{m} P false(C p_{j} false| p_{i} false), j = 1, 2,

s.t. c_{i} = 0 .

The necessary number of encoded packets to recuperate initial information, with K input symbols used in LT encoding, is

N = K + scriptO false(\sqrt{K} \log^{2} K false) .

…”

Section: System Model and Problem Formulationmentioning

confidence: 99%

Exploiting classification for fountain data estimation in wireless sensor networks

Belabed

Bouallègue

2020

Int J Communication

View full text Add to dashboard Cite

show abstract

“…The RSD exhibits signification improvement in performance over ISD. However, the RSD is not the optimal degree distribution for finite length LT codes [7]. Poisson RSD (PRSD) was designed by combining Poisson distribution (PD) and RSD to reduce average overhead and improve coding efficiency for LT codes [8].…”

Section: Introductionmentioning

confidence: 99%

Goal‐oriented design of optimal degree distribution for LT codes

Zhang

2020

IET Communications

View full text Add to dashboard Cite

“…LT codes were originally designed over binary erasure channels (BEC) with two decoding methods, namely, belief propagation (BP) and Gaussian elimination (GE). BP is a fast algorithm with complexity of O(klog(k)), and its performance of successful decoding is exploited in [10]. GE is a maximum likelihood decoding with complexity of O(k 2 ), which was further developed by Kim et al [11], [12] and Bioglio et al [13].…”

Section: Introductionmentioning

confidence: 99%

A Greedy Spreading Serial Decoding of LT Codes

Lei

Huang

2019

IEEE Access

View full text Add to dashboard Cite

In this paper, we propose new serial decoding of Luby transform (LT) codes over additive white Gaussian noise channels. LT encoder generates a potentially limitless number of encoded packets, and the decoder incrementally collects the packets to ensure successful recovery of the information. In the proposed algorithm, the newly coming code nodes are the first to pass the messages, and it is their neighboring source nodes that will receive these input messages and update their output messages; then, the next neighboring code nodes that have not been covered are to be included in the updating group; in this greedy way, the message propagation is conducted from neighbors to neighbors. The analysis demonstrates that the proposed algorithm has a faster convergence speed than the conventional ones, and simulation shows that it has an effective bit error rate performance. INDEX TERMS AWGN channel, fountain codes, LT codes, soft decoding.

show abstract

Performance Analysis for Finite Length LT Codes via Classical Probability Evaluation

Cited by 11 publications

References 16 publications

Exploiting classification for fountain data estimation in wireless sensor networks

Exploiting classification for fountain data estimation in wireless sensor networks

Goal‐oriented design of optimal degree distribution for LT codes

A Greedy Spreading Serial Decoding of LT Codes

Contact Info

Product

Resources

About