Reliable distortionless jpeg image transmission over Mimo system using antenna selection.

Deepa, R. ; Baskaran, K.

Introduction

Of all communication services available today, wireless services are having a dramatic impact on our personal and professional lives. In real-time wireless environment, the signal propagates from the transmitter to the receiver through multipaths providing a form of MIMO system. JPEG Standard [1] proposed in 1992 is widely used for still image compression and standard. W.B. Pennebaker and J.L Mitchell introduced the JPEG still image standard [2]. In literature joint source channel coding has been discussed in [3-5]. Linear receivers have been discussed in [6-8] and Alamouti introduced the concept of MIMO system [9]. Sabir et. al has done extensive research on unequal power allocation scheme for JPEG transmission[10-11].

In this paper, we provide an efficient scheme for transmission of Joint Photograph Experts Group (JPEG) compressed images over MIMO system by employing spatial multiplexing. The image under test is compressed using JPEG compression algorithm and all the pixels are transmitted with equal power. We use V-BLAST ZF receiver for symbol detection and the image is reconstructed by decompression algorithm at the receiver. The paper is organized as follows: Section II introduces the system model and power allocation scheme in Section III. Section IV describes the symbol detection scheme for estimating the MIMO channel input and section V provides the simulation details. In section VI, we discuss our results and future enhancements.

System Model

[FIGURE 1 OMITTED]

Source Coder

A progressive Discrete Cosine Transform (DCT) based JPEG coder with spectral selection mode is used. DCT-II is performed over small 8x8 blocks according to the equation (1) given below.

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

The DCT coefficients are organized into 64 quality layers: 1 dc layer followed by 63 ac layers. The resolution and quality of the reconstructed image improve when more layers are decoded.

Headers and Markers

Within each layer, headers and reset markers are introduced to prevent error propagation between different parts of the bit stream. Bit errors occurring during transmission can affect the headers, the markers, the dc layer and ac layers. If there is an error in the header, the entire image will be damaged and cannot be recovered. In case of error in reset markers, synchronization will be lost. So it is assumed that headers and reset markers are transmitted error free.

Huffman Coding

Huffman coding is a statistical technique that attempts to reduce the amount of bits by encoding most frequently occurring symbols with shorter codes and longer codes for less significant symbols.

[FIGURE 2 OMITTED]

POWER ALLOCATION SCHEME

The basic scheme used for transmission over MIMO system is equal power allocation. In this scheme, the data is divided into smaller blocks corresponding to the number of antenna used in the system and the total power is split up equally to each block for transmission.

[FIGURE 3 OMITTED]

Antenna Selection

The number of transmitters and receivers considered here is 4. We initially transmit the same set of data through all the four antennas and determine the mean squared error (MSE) of the received data for various paths. Mathematically SINR can be found using the relation (2), assuming the channel is well known to the transmitter and receiver.

[[epsilon].sub.k] = 1/[[([X.sub.n]H*H+[I.sub.k]).sup.-1]]-1 (2)

The received data 'y' is given by

Y = a x H + noice (3)

Where 'a' is the transmitted vector and H is the random channel. The noise added is AWGN with zero mean and power spectral density No/2. We calculate the SINR of various paths and select the transmit antenna corresponding to the highest SINR to transmit the most important stream like edges, the transmit antenna with the next highest SINR to transmit the other important feature and so on.

Symbol Detection

It estimates the MIMO channel input 'a' from the received vector 'y' assuming that the receiver has perfect knowledge about the channel (H). The channel considered is a random channel and is assumed to be constant for the entire duration of a frame.

Zero-Forcing receiver

ZF is a low-complexity linear detection algorithm that produces

a = Q([a.sub.zf])

where [a.sub.zf] = H*xY

where H* denotes the Moore-Penrose pseudo inverse of H, which is a generalized inverse that exists when H is rank deficient. The ZF receiver eliminates co-channel interference entirely since H*xH = 1.

Simulation Details

A database of 25 grayscale and color images of random size is used for simulation. First we check whether the image is a grayscale or color image. The color image is converted from RGB space to YCbCr and DCT is performed over small 8x8 blocks of the image. We assume the number of transmitters and receivers to be 4 and a random channel H, which is perfectly known to the transmitter and the receiver. The channel is assumed to be constant for the entire duration of a frame. 4-QAM modulation is performed over all blocks. We initially transmit the same set of data through all the four antennas and determine the mean squared error (MSE) of the received data for various paths. The SINR of various paths are calculated and the transmit antenna corresponding to the highest SINR is selected to transmit the most important stream like edges, the transmit antenna with the next highest SINR to transmit the other important feature and so on. V-BLAST ZF algorithm is used to estimate the transmitted vector and the estimation is on a symbol-by-symbol basis.

Results and Discussion

The digital transmission of image over a noisy channel can be improved by selecting the best antenna. The transmit antenna corresponding to the highest SINR path is selected as the best antenna. The performance of this scheme is validated by computing its symbol error rate (SER). The symbol here represents 4-QAM modulated data and has 2 bits. The SER is reduced considerably for transmission using antenna selection. For an Eb/No of 10 dB, the SER is reduced by 25% if antenna selection is performed. In image transmission using antenna selection, the very important streams have less distortion compared to equal power allocation with no antenna selection. Original Lena image (Figure 4) and reconstructed images (Figure 5 and 6) are given. BER and SER for both transmission of image with and without antenna selection are given in Figure 7 and Figure 8. The above results prove that transmission of important feature of the image through the best antenna outperforms the known scheme.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

Conclusion

The key advantages of MIMO are exploited to obtain a reliable distortionless image transmission. Not all parts of the image are important and there can be some compensation in those parts to achieve less error in the significant part of the image. Instead of transmitting the pixels with equal power through all transmit antennas; the path with highest SINR is selected to transmit the most important feature of the image and the next best antenna to transmit the other important feature and so on. Results show that the proposed scheme provides significant image quality improvement and less distortion compared with equal power allocation with no antenna selection.

References

[1] ISO/IEC JTC1, "Digital compression and coding for continuous-tone still images part 1: Requirements and guidelines," JPEG-9-R6 CD10918-1, 1991.

[2] W.B.Pennebaker and J.L.Mitchell, "JPEG still image data compression standard", Van Nostrand Reinhold, 1993.

[3] Y. Eisenberg, C.Luna, T.Pappas, R.Berry and A.Katsaggelos, "Joint source coding and transmission power management for energy efficient wireless video communications", IEEE transactions--Circuit Syst. Video technology, vol.12, pp 411-424, June 2002.

[4] I. Kozintsev and K.Ramchandran, "Robust image transmission over energy constrained time-varying channels using multiresolution joint source channel coding", IEEE Trans. Signal Processing, vol. 46, pp.1012-26, April 1998.

[5] Maurizio Murroni, "A power-based unequal error protection system for digital cinema broadcasting over wireless channels", Science direct, Sig. Proc. Image Communication, vol 22, pp. 331-339, 2007.

[6] M.F.Sabir, H.R.Sheikh, R.W. Heath Jr. and A.C. Bovik, "A joint source-channel distortion model for JPEG compressed images", IEEE Trans. Image Processing.

[7] M.F.Sabir, R.W. Heath Jr. and A.C. Bovik, "Unequal power allocation for JPEG transmission over MIMO systems", IEEE publications 2005.

[8] S.M. Alamouti, "A simple transmit diversity technique for wireless communications," IEEE jour. Select. Areas Commn.,vol 16,pp 1451-58, Oct 1998.

[9] A.Paulraj, R.Nabar and D.Gore, "Introduction to space-time wireless communications", Cambridge University Press, 2003.

[10] M.K.Simon and M.S.Alouni, "Digital Communication over fading channel", John Wiley and Sons, 2000.

[11] Hamid Jafarkhani, "Space time coding--Theory and practice", Cambridge University Press, 2003.

R. Deepa (1) * and K. Baskaran (2)

(1) Senior Lectrer, ECE, Amrita Vishwa Vidyapeetham, Coimbatore, Tamil Nadu, India.

(2) Senior Lecturer, CSE, Government College of Technology, Coimbatore, Tamil Nadu, India.

* Corresponding author E-mail: [email protected]