Beamforming System Architectures
Beamforming architectures are classified into three categories: analog beamforming, digital beamforming, and hybrid beamforming architectures. Analog beamforming This architecture is also known as phased arrays in the field of radar and antennas. It uses attenuators and phase shifters as part of the analog RF circuit where a single data stream is divided into separate paths. The advantage of this method is that it only requires a single digital and RF chain. The disadvantage is that a single data stream can be transmitted at a time with a given weighting vector. Fig. 2.6 illustrates a basic architecture of an analog beamforming transmitter, it is made up of a single RF chain and several phase shifters feeding an antenna array.
Data stream RF processing Digital baseband DAC LPF processing LO SRF Phase shifter + PA Figure 2.6: Analog beamforming architecture  Analog beamforming can be carried out at either the RF or intermediate frequency level . The weighting vector w is achieved by varying the amplitude and phase of the signals being fed into the antenna array elements.
Digital Beamforming Digital beamforming usually use as many RF chains as the number of antenna phase weighting are applied. This is typically the preferred approach at sub 6 GHz frequencies as the the RF chain components are comparatively inexpensive and it offers greater flexibility for the signal processing. Fig. 2.7 illustrates the high-level digital beamforming transmitter architecture with multiple RF chains . Using this architecture, a large number of data streams can be sent with unique weighting vectors, depending on the number of antennas and the propagation channel. This enables addressing multiple users or location with different data sharing the same resources, i.e., Spatial Division Multiplexing (SDM) and/or increasing the data rate of a communication by multiplexing the data to be transmitted over several orthog-onal channels, i.e., Spatial Multiplexing (SM), if the user is also equipped with mul-tiple antennas. Digital beamforming can also benefit from the array gain if the N RF chains are coherent.
Hybrid beamforming is a trade-off between digital and analog beamforming archi-tectures. Similar to digital approaches, this allows for flexibility in the processing (with the possibility to address multiple users and/or do SM) while reducing the overall cost and power consumption of the array thanks to the use of analog beam-formers . The idea is to have less digital and RF chains than the number of antenna elements. It is characterized by separate analog beamforming unit con-nected to a sub-set of antennas for each data stream. This approach is preferred in the millimeter-wave domain where the number of antenna elements in the array is very large and the hardware power consumption too. Analog beamforming unit losses due to phase shifters can be mitigated by replacing the adaptive phase shifters with a selective beamformer such as a Butler matrix. Fig. 2.8 shows a typical hybrid beamforming transmitter architecture. The pre-coding is divided between the ana-log and digital domains. In this example, K << N data streams can benefit from a unique weighting vector w. Signals at the output of each RF chain are sent over a sub-array and can experience an array gain of N/K. The number K of data streams is equal or less than the number of RF chains, depending on the considered precoding algorithm.
Beamforming in Geocasting
In the literature, beamforming has been considered in geocasting applications in the area of vehicular communication, a typical case is in the car-to-X project mentioned in chapter 1. The introduction of beamforming enhanced the application of geo-casting both in the cases of multi-hop implementation, i.e., when vehicles need to forward messages to other vehicles, and in the Infrastructure to vehicle communi-cation (I2V), i.e., when road side units (RSU) need to communicate to vehicles. For simplicity and in line with our work, the following discussion will only consider the typical I2V implementation where the RSU sends information to the vehicle nodes. It was observed in the car-to-X project that using quasi-omnidirectional antennas made the geocasting scheme vulnerable to interference and lead to loss of power with respect to the position of the targeted receiver. To mitigate against the above, multiple schemes [53, 54, 55], were initially proposed to attain directivity. In , C2X beamforming is proposed to secure the communication, enable user privacy, and congestion control, all carried out at the physical layer. To do so, the scheme employed radiation pattern control mechanisms that allowed nodes to focus trans-mitted power in a desired direction through adaptive beamforming, this enabled the base stations to improve both signal reception and rejection of unwanted sig-nals . The approach ensures that the data being exchanged is guarded and the medium on which an attacker can operate is restricted . To do so, a unique X-Y (two crossed) antenna structure capable of carrying out both broadside and endfire transmission was designed to comply with the minimum antenna requirements for the C2X communication defined in .
At minimum, the requirements set that transmitted packets at any instant had a geographical validity as a function of receiver position with respect to the base station. The assumption here being that directing the radiation pattern based on its geographical validity limits the reach of an attacker by excluding its position from the transmission path. It was also a requirement that in cases of Denial-of-Service attack, where it is an attacker that is transmitting, the antenna system had to imple-ment null steering in the direction of the attacker/jammer, again the assumption was that in most cases the attacking system will be located on the road side. The design flow and algorithms used to implement phasing, element biasing, or transmission power was determined based on given scenarios, like attacker position and message type, at any given instant as described in .
Beamforming Algorithm: Secure Transmission
In C2X beamforming, selection and steering of the beams is carried out at the lower system abstraction layers of the architecture. This section briefly describes a typical beamforming algorithm whenever a Decentralized Environment Notification Mes-sage (DENM) is to be transmitted. The DENM is initially passed from the appli-cation layer for subsequent transmission. This messages header is analyzed by a beamforming component of the MAC layer, so as to determine the communication scenario linked to it. Assuming the originator field determines that the message is for broadcasting, antenna elements x6 and x7 will be biased for a near omnidi-rectional pattern, here suggesting that the message is to be received by all nodes around the transmitting node. In a scenario where it is only required to forward the given message, then an endfire pattern is created. To do so, specific elements of the Y-antenna array are biased with the appropriate electronic phase shift (here 207 ). Simultaneously, the power level of the feeding signal can be varied based on the required communication range. There is no rule of thumb as to the relationship of power in dB with possible range, say in meters, as this scheme is expected to operate in diverse environments which always experience different channels. Specific to C2X beamforming, probe Vehicle Data is sent every time the vehicle passes a RSU. In that case, the position-dependent pattern to produce quite narrow beams towards the receiver node. Assuming non stationary vehicles, the process of steering the beam relative to the dynamic positions of the vehicle and related geometric positioning between the two nodes is calculated and transferred to an electrical phase shift of the feeding signal. Indeed Mutual beamforming between nodes results in significantly improved directivity, at the same time isolating poten-tial eavesdropping of the exchanged messages.
Metrics for analysing DM Systems perfomance
Multiple proposition have been made for the acceptable metric upon which to ana-lyze DM performance. One such metric is the normalized error rate adopted in [73, 77]. However this approach did not account for magnitude and phase reference of detected constellation patterns were defined. Also having not considered channel noise and coding approach, this metric is not able to account for differences in per-formance given scenarios where: A constellation symbol is constrained within its optimal compartment decoding, say different locations with a QPSK quadrant; A constellation falls into a different compartment all together. With this omission, a DM systems ’error rate’ is thus challenging to use for systematic analysis.
In  an error-vector-magnitude-like figure of merit, that defines the capability of constellation patter distortion in DM systems is defined. This scheme faces the same challenges of lack of channel noise and coding strategy consideration. In ad-dition, it was noted that its computation does generate distinct values for static and dynamic systems because they employ different constellation patterns.
In , i.e., the dual beam beam directional modulation earlier mentioned, it is the bit error rate (BER) that is employed to analyse the performance of the QPSK DM system proposed. The closed-form QPSK BER lower bound equation for static DM scheme evaluation was proposed as the metric of choice for [67, 66].
In  it was demonstrated that BER computed from closed from equations or transmitted data streams together with secrecy rate was adequate metrics for evalu-ating static DM systems. Secrecy rate being a metric already in wide use in the infor-mation theory and specifically the cryptography field. It was equally demonstrated that for dynamic DM systems that experience zero-mean Gaussian distributed or-thogonal interference, EVM-like metrics, BER, and secrecy rate exhibited similar re-sults that could as well be inter-converted.
Conclusions on Directional Modulation
While DM can address the issue faced by beamforming of having a ZOR size that depends upon the received SNR, and hence upon the user hardware and conditions, it however does not overcome the limitation introduced by the inversely propor-tional relationship between array aperture size and beamwidth. Indeed, its focusing performance, i.e., its angular selectivity, highly depends of the environment, and for most DM techniques in a LOS case, are not better than beamforming.
Table of contents :
1.1.1 Geocasting concept
1.1.2 Geocasting in Wireless Communications
1.2 Objectives of the PhD
1.3 Thesis outline
2 Spatial Focusing approaches: State of the art
2.1 Antenna Array Theory
2.2.1 Linear array
2.2.2 Beamforming System Architectures
2.2.3 Beamforming in Geocasting
Antenna Model Specification
Beamforming Algorithm: Secure Transmission
2.2.4 Conclusion on beamforming
2.3 Directional Modulation
2.3.1 Metrics for analysing DM Systems perfomance
2.3.2 Conclusions on Directional Modulation
2.4 Time reversal
2.4.1 Conclusions on Time Reversal
3 Spatial Data Focusing (SDF)
3.1 Spatial Data Focusing framework
3.2 Orthogonal Signaling
3.2.2 Orthogonal signaling design approaches
Spread spectrum orthogonal coding
3.3 Spread Spectrum Modulation
3.3.1 Classification of spreading codes
Orthogonal vs. non-orthogonal codes
Real vs. complex codes
Analog vs. digital codes
3.3.2 Pseudo-Noise (PN)
Code Generation by Linear Feedback Shift Registers
3.3.3 Orthogonal codes
Walsh Hadarmard (WH) Codes
Orthogonal Gold Codes
3.3.4 Correlation characteristics of spreading codes
3.3.5 Direct Sequence Spread Spectrum (DSSS)
3.4 Principle of DSSS based Spatial Data Focusing
3.4.2 Choice of spreading codes
Orthogonal Gold codes
3.4.3 Channel model
Rx-Tx signal phase difference measurements
Mechanism responsible for the directional behaviour of the data transmission
Validation of the symbolic and analytical DSSS-SDF realizations
Validation of Spatial Selectivity: A comparison of SDF with
Influence of spreading sequences on the robustness of the scheme
3.5 DSSS-based SDF with IQ resources in LOS
3.5.1 Principal of IQSS-SDF
3.5.2 Matched filter based receiver
Channel estimation and equalization
Role of Spreading sequences
Influence of the channel estimation on angular selectivity .
Beamwidth as a function of N
Effect of b (inter-element spacing)
SDF operation robustness to noise
4 Influence of multipath channels on DSSS-SDF-IQ
4.2 Multipath resolution capacity of classical Direct Sequence Spread Spectrum (DSSS)
4.3 DSSS-SDF-IQ over multipath channels: A general case
4.4 Characterizing wireless channel via Ray Tracing
4.4.1 Power delay profile (PDP)
4.4.2 RMS angle spread
4.4.4 RMS Delay Spread
4.5 DSSS-SDF-IQ over urban canyon channel model
4.5.1 Specular Multipath Modelling
4.5.2 Two-rays Ground-Reflection channel model
BER vs. Receiver angular orientation
4.5.3 4-ray channel model: urban canyon
Characterizing the 4-ray urban canyon channel model
4.5.4 6-ray channel model: urban canyon
Simulation and Results
Urban canyon 6-ray channel model characterization
Direct Sequence Spread Spectrum – Spatial Data Focusing with
IQ (DSSS-SDF-IQ) simulation results
A practical demonstration of the robustness of Direct Sequence
Spread Spectrum – Spatial Data Focusing with IQ (DSSS-SDF-IQ)
4.6 The influence of space-time geometrical channel models on SDF
4.6.1 Geometrically Based Single Bounce Macrocell Channel Model .
Geometry of the Geometrically Based Single Bounce Macrocell
Geometrically Based Single Bounce Macrocell (GBSBM) channel
4.6.2 Perspective and Conclusion
5 Conclusions and perspective
List of publications