Graham Ross Daniels

Transmitter Noise Effect on the Performance of a MIMO-OFDM Hardware Implementation Achieving Improved Coverage

This paper presents analysis of performance measurements from a MIMO-OFDM IEEE 802.11n hardware implementation at 5.2 GHz using four transmitters and four receivers. Two spatial multiplexing systems are compared; one which uses a zero-forcing (ZF) detector and the other a list sphere detector (LSD). We show that the measured results do not align with standard prediction based on simulation assuming uncorrelated receiver noise. We show that the discrepancy can be explained by the inclusion of transmitter noise into the channel model. This effect is not included in existing MIMO-OFDM channel models. The measured results from our hardware implementation show successful packet transmission at 600 Mb/s with 15 bits/s/Hz spectral efficiency at 73% coverage for ZF and 84% coverage for LSD with an average receiver signal to noise ratio (SNR) of 26 dB.

Performance of MIMO-OFDM-BICM on Measured Indoor Channels

In this paper, the bit error rate (BER) and packet error rate (PER) performance of multiple-input multiple-output orthogonal frequency division multiplexing with bit interleaved coded modulation (MIMO-OFDM-BICM) systems using convolutional, turbo, or low density parity check (LDPC) coding schemes are evaluated on MIMO-OFDM channels measured in various indoor environments. For a given average signal to noise ratio (SNR) per receiver, the performance of MIMO-OFDM-BICM in a line-of-sight (LoS) environment can vary significantly depending on the existence of scattering, while it is more uniformly distributed in a non-line-of-sight (NLoS) environment. We show that, due to bursty coded bit error characteristics of turbo and LDPC coding, only marginal coding gain over convolutional coding is observed in terms of BER. However, when PER is compared, a 3 dB coding gain is achieved by the turbo or LDPC code over the convolutional code

Design Criteria of MIMO Systems

In this paper, the uncoded symbol error rate (SER) versus signal to noise ratio (SNR) curves of M-ary quadrature amplitude modulation (M-QAM) multiple-input multiple-output (MIMO) systems using zero-forcing (ZF) or maximum likelihood (ML) detection employing various numbers of antennas are given analytically and empirically. Using the given formulas, the designer of a MIMO system can easily analyse the tradeoff between using ZF or MLD, where ZF is computationally efficient but requires more receivers to achieve the same performance as MLD. An example is given, showing that, for an uncoded MIMO system using QPSK, 16QAM, or 64QAM, to achieve the same or better performance as MLD using 2, 3, and 4 antennas at each end, ZF requires at least 3, 5, and 7 receivers, respectively, at a specified average SER of 10-3

Practical performance of MIMO-OFDM-LDPC with low complexity double iterative receiver

This paper considers MIMO-OFDM transmission with low density parity check (LDPC) codes. We employ a low complexity minimum mean-square-error (MMSE) soft-interference-cancelation (SIC) based double iterative receiver (DIR). Results are presented for a real system implementation at 5.2 GHz. We achieve zero packet errors at 90% of the measured indoor locations, when transmitting at 600 Mbit/s, with 15 bit/s/Hz spectral efficiency and 26 dB signal-to-noise ratio. We show that the proposed receiver actually outperforms a list sphere detection (LSD) based single iterative receiver (SIR) at high coding rates, in practice. Investigations reveal that the LSDSIR is adversely affected by the non-Gaussian noise present at the receiver, while at the same time the MMSE-SIC-DIR is better able to handle transmitter noise present in the practical system.

Computational efficiency of list sphere detection

In this paper the performance of a promising MIMO detection algorithm, list sphere detection, is evaluated for a range of list lengths and for four variations to the basic algorithm. The evaluation uses a MIMO-OFDM-BICM system simulation to show bit error performance versus the computational effort, and the results are compared to maximum likelihood detection. The results provide guidance on the selection of an appropriate list length and algorithm. In applications where a short list length must be used due to memory or computation constraints the results show the loss in performance that can be expected