This paper presents a cross-layer optimization method for MAC protocol data unit aggregation, abbreviated to A-MPDU, with high throughput (HT) immediate block acknowledgement in non-competitive channel environment. We derive the goodput as a function of the transmission strategy, including the aggregated number of MPDUs and length of each MPDU of the MAC layer, and modulation and channel coding scheme (MCS) of the PHY layer. Among all the possible transmission strategies complying with packet error probability (PER) requirement, the one with the highest goodput can be selected. The simulation result shows that the system's goodput attained by the proposed method is twice as large as the one without A-MPDU.

This paper proposes an RST-invariant multi-bit watermarking system based on LPM and DFT. This system embeds the watermark in a rotation and translation invariant domain obtained by first acquiring log-polar mapped DFT (LPM-DFT) magnitudes of an image, and then taking the magnitude of the 1-dimensional DFT along the angular axis of these LPM-DFT magnitudes. Neither the original cover image nor a presumed watermark is required during watermark extraction. Scaling invariant is accomplished by a simple search along the log-radius axis. Simulation results indicate that this new system has high robustness against RST distortion and moderate robustness against image compression such as JPEG.

We derive expressions for the symbol error probabilities for a recording code concatenated with a Reed-Solomon (RS) code. The recording code is structured by a guided scrambling (GS) code for the direct current (DC) suppression in conjunction with a runlength-limited (RLL) code. As for the GS codes, convolutional GS and Galois field (GF) addition GS schemes are examined. As for the RLL codes, two types of RLL codes are investigated. One is a traditional RLL code where a bit length m of an RS symbol is an integer multiple of a bit length p of an RLL source symbol. The other is a new type of high-rate RLL code where p>m. We compute the RS symbol error rates when these RLL codes are combined with the two GS schemes.

Guided scrambling (GS) is used to control the runlength within codewords as well as to suppress dc components. A weakly constrained code designed by using a GS technique violates the imposed /spl kappa/-constraint with low probability, but achieves a code rate close to the channel capacity. In this study, the probability that codewords violate the k-constraint was measured by using convolutional GS and by using Galois-Field (GF)-addition GS. The occurrence probability of each runlength was also computed and then compared to a 24/25(0,8) block code having a high code rate and to the maxentropic runlength limited (RLL) sequence. The comparison shows that codes designed by these GS techniques attain extremely high code efficiency.