In this paper, we study the capacity and symbol error probability (SEP) of generalized spatial modulation (GSM) multiple-input multiple-output (MIMO) using measured channels that are obtained by channel sounding in an indoor office environment at 60GHz. Spatial modulation (SM) and GSM are emerging low-complexity MIMO schemes that have been extensively researched for low-GHz (below 6GHz) communications. Recently, they have been considered and shown to be promising also for (mmWave) communications. In the simplest possible case, they require only one RF chain both at the transmitter (TX) and receiver (RX), and thus, are especially attractive for mmWave communications, in which the number of RF chains needs to be as low as possible. Despite of some early works on the theoretical analysis of SM/GSM for mmWave communications, there have been no investigations using real-world channel data. We focus on the office line-of-sight (LOS) scenario and investigate three problems: 1) the performance of GSM using the extracted LOS component of measured channels; 2) the impact of non-LOS NLOS components on the performance of GSM; and 3) possible simple modulation and reception algorithms for GSM that rely only on the LOS component of the channel. The results being reported in this paper not only validate the main claims of previous studies based on ideal pure LOS channels, but also lead to novel findings. One major conclusion is that NLOS components are harmful to the SEP of GSM and should be avoided. As another important outcome, our results strongly motivate the use of precoding in GSM systems to simultaneously improve the channel capacity and reduce the physical size of MIMO arrays (thus eliminating one major issue of LOS GSM).

Spatial modulation (SM) and generalized spatial modulation (GSM) are emerging multiple input multiple output (MIMO) schemes that use transmitter (TX) antenna switching for data transmission. Their operating principle makes optimization of channel capacity and mutual information usually more difficult than for conventional MIMO schemes which are not based on antenna switching. We propose to use channel cut-off rate as a relevant and more tractable metric for performance optimization of spatial modulation (SM)/GSM systems, as it constitutes a practical lower-bound of channel capacity. In particular, we propose four TX precoding schemes for increasing the cut-off rate of SM/GSM systems. We show that those TX precoding schemes which are designed for increasing array gain provide the largest improvement of cut-off rate for low signal-to-noise ratio (SNR). On the other hand, the TX precoding schemes that are designed for increasing the minimum Euclidean distance of GSM symbols are more suitable for application to medium to high SNR setups and correlated channels. The proposed precoding schemes are shown to be able to enhance mutual information, and the gain is shown to be of the same order of magnitude as the gain of the corresponding channel cut-off rate.

In this paper, we propose to use variable-Nugeneralized spatial modulation (VGSM), which is a specific type of generalized spatial modulation, in indoor line-of-sight (LOS) millimeter-wave (mmWave) communication. As compared with fixed-Nugeneralized spatial modulation, VGSM needs less number of transmitter (TX) antennas to achieve the same data rate. Reducing the number of antennas is especially important for LOS mmWave MIMO communication, since it means reduced array lengths and routing loss in the RF front-end. We first derive and analyze the channel capacity for LOS VGSM, and then propose a novel switching structure for implementation of the VGSM TX at mmWave frequencies and analyze its performance in terms of power added efficiency and capacity. Two optimizations of the system performance of VGSM are performed: 1) power allocation optimization for the proposed TX, and 2) antenna separation optimization in order to optimize the capacity in LOS. Our results show that VGSM is a feasible and promising scheme for indoor LOS mmWave communication. For example, we show that an 8 × 8 VGSM system with practical component values can achieve more than 20bpcu spectral efficiency within a distance of 1-5m and more than 17bpcu within 5-10m.

In this paper, we propose to use spatial modulation (SM) and the generalized spatial modulation (GSM) MIMO schemes in indoor line-of-sight (LOS) millimeter-wave (mmWave) communication at 60GHz. SM/GSM are known but only at low-GHz frequencies where the channels are typically rich scattered and characterized by Rayleigh or Rician distribution. However, 60-GHz indoor channels are typically not rich but rather sparsely scattered and dominated by the LOS component, thus making them different from low-GHz fading channels and our work novel. We first seek to optimize SM in LOS by finding the channel conditions that minimize its symbol error probability (SEP). Then, we extend our studies to LOS GSM and derive its channel capacity and SEP. Furthermore, we present numerical studies on the behavior and performance of SM/GSM in LOS. LOS spatial multiplexing and beamforming MIMO schemes are used as benchmarks for comparison. At last, we propose novel TX and RX hardware architectures, both of which use only a single RF chain, for implementation of SM/GSM at mmWave frequencies. Both simplicity and good performance are exhibited by LOS SM/GSM thus making them very attractive techniques for short-range indoor mmWave communications at 60GHz.

In this paper Minimum Mean Square Error (MMSE) precoding for the Preprocessing aided Spatial Modulation (PSM) transmission scheme is investigated. PSM is capable to transmit data both in the spatial domain and in the modulation domain. We derive the upper bound for the Average Bit Error Probability (ABEP) for MMSE precoded PSM data transmission and we show by means of simulations that the bound is tight for a large signal to noise ratio (SNR) range. Also, the influence of the number of transmit and receive antennas on the ABEP is analyzed. It is shown that the increase of the number of transmit antennas will improve the ABEP. We demonstrated that the analytical ABEP for MMSE precoded PSM is considerably lower than for Zero-Forcing (ZF) precoded PSM. We determined that for constant bit rate, the lowest Bit Error Rate (BER) is achieved with a balanced number of bits in the spatial and modulation domain. The correlation at the receive antennas has a dominant influence on the BER.

We propose a novel diversity combining method called local average gain combining (LAGC) that is based on the local average of channel gains. Compared with MRC, which requires the instantaneous complex channel coefficients, and EGC, which requires only the instantaneous channel phases, LAGC requires the instantaneous channel phases but the lo- cal average of the channel gains, implying an implementation complexity between those of MRC and EGC. The average bit error probability (ABEP) of LAGC is analyzed and derived in closed-form using a method that is based on the characteristic function (cf) and the Paserval's theorem, showing that LAGC approximates to MRC in slow fading and to EGC in balanced fast fading. Simulations are performed to validate the analysis. The channel phase estimation error is included in the simulation, and it turns out that LAGC and EGC significantly outperform MRC and SC. This is a distinct advantage of LAGC because in fast fading it is more difficult to track the channel coefficients and on average the channel phase error is larger.

Receive antenna diversity is a technique that uses multiple antennas at the receiver side to improve signal reception performance by exploiting spatial diversity. However, in a mobile diversity receiver, the diversity performance is degraded by mobile antenna mismatch created by, e.g., body and hand effects. In this paper, we first explain the mechanism of antenna mismatch and then derive analytic results on the distribution and mean value of the combined SNR in a general frequency-flat Rayleigh fading environment. Three combing techniques are considered: MRC (maximal ratio combing), EGC (equal gain combing) and SC (selection combing). We also present simulated results on BER performance showing that, in a mobile terminal with two receive antennas, single mobile antenna mismatch up to -7 dB causes considerable degradation in array gain, but only negligible reduction in diversity order such that the diversity system still exhibits significant improvement in BER performance compared to an ideal single receive antenna system. Even when both antennas have a mismatch up to -7 dB, the diversity system shows noticeable improvement in BER performance at high SNR.