Robust JPEG image transmission using unequal error protection and code combining

Circuits Syst Signal Process

2010

Self Cite

In this paper, we combine two Joint Source Channel Decoding (JSCD) approaches and apply them to progressive JPEG image transmission. The first JSCD approach consists of optimizing the allocation of the channel code-rates to the different layers of the JPEG image, and is commonly referred as Unequal Error Protection (UEP). The second JSCD strategy exploits a priori probabilities obtained from source statistics as well as the residual redundancy of the source to improve channel decoding. Additionally, we propose an error resilient sub-categorization scheme for the DC layer so as to reduce the effect of the channel-induced distortion. The hybrid JSCD scheme is implemented with Rate Compatible Punctured Turbo Codes and also with the bandwidth efficient Rate Compatible Punctured Turbo Trellis Coded Modulation. Gains of over 10 dB in PSNR are obtained with the hybrid JSCD scheme as compared to a conventional JPEG image transmission scheme.

“…In [4], we modified the distortion model of [13] and obtained the worse case total expected distortion for a JPEG image that has been split into 5 sub-bands as follows:…”

Section: Optimization and Distortion Modelingmentioning

confidence: 99%

“…When the sub-bands are divided into blocks of size B, the total distortion on a per image block basis is obtained as follows [4]:…”

Section: Optimization and Distortion Modelingmentioning

confidence: 99%

mentioning

confidence: 99%

See 1 more Smart Citation

Hybrid Joint Source Channel Decoding for Progressive JPEG Image Transmission

Circuits Syst Signal Process

2010

Self Cite

“…Finally the third one, (UEP3), only optimizes the allocation of retransmissions by using the bandwidth allocated for retransmission requests. In [10] we only compared the performance of a single UEP scheme employing RCPC codes and code-combining with its EEP counterpart.…”

mentioning

confidence: 99%

“…Furthermore in [10], we proposed a robust JPEG image transmission scheme using UEP with Rate Compatible Punctured Convolutional Codes (RCPC) and Code-Combining, in which a new technique for jointly optimizing the allocation of channel code-rates and number of repeats was developed. The scheme we developed in [10] did not require a feedback channel as the retransmissions were sent a-priori.…”

mentioning

confidence: 99%

Highly scalable JPEG image transmission with unequal error protection and optimal feedback

2010

Telecommun Syst

Self Cite

This paper investigates the performance of three different Unequal Error Protection (UEP) schemes for progressive JPEG image transmission using delay-constrained hybrid ARQ, with iterative bit and symbol combining. The first UEP scheme considers only the optimization of channel code-rates and keeps the number of retransmissions fixed for all the subbands of the image. The second one optimizes both the channel code-rates and retransmissions, while the third only considers the optimal allocation of retransmission requests. The UEP schemes are designed with two different coding techniques. The first one employs Rate Compatible Punctured Turbo Codes (RCPT) with iterative bit combining and, is suitable for applications requiring high power efficiency. For the second one we propose a new coding strategy, Rate Compatible Punctured Turbo Trellis Coded Modulation (RCPTTCM) with iterative symbol combining, which provides high scalability and bandwidth efficiency. Gains of over 9 dB in Peak-Signal-to-Noise-Ratio are obtained with the UEP schemes as compared to their corresponding Equal Error Protection (EEP) schemes.

Wireless JPEG Image Transmission Using Multiple Diversity Combining and Unequal Error Protection

2010

Wireless Pers Commun

Self Cite

In this paper we investigate the performance of JPEG image transmission over the IEEE 802.11a WLAN, using Multiple Diversity Combining (MDC) and Unequal Error Protection (UEP). An MDC scheme is obtained by simultaneously employing transmitter diversity, receiver diversity and retransmission diversity. UEP is performed by first optimizing the number of headers allocated to the subbands of a JPEG image to prevent error propagation, and secondly by optimizing the number of retransmissions with a view to minimize the channel induced distortion. The MDC-UEP strategy has been employed with both Convolutional and Turbo codes. Simulation results show that a gain of 8 dB in PSNR can be obtained with MDC-UEP in the range 0 dB ≤ Eb/No ≤ 3 dB, over a scheme which uses MDC with Equal Error Protection (EEP). Moreover MDC-UEP with Turbo codes provides a gain of 0.5 dB in PSNR and 5 percent in throughput over an MDC-UEP scheme with Convolutional codes.