Distributed Compression and Transmission with Energy Harvesting Sensors

Get Complete Project Material File(s) Now! »

Multi-antenna Systems with Imperfect CSI

It is now well known that channel adaptation and precoding techniques in multi-antenna systems provide large improvements in the throughput [37]. These techniques require some knowledge of the wireless channel conditions also known as CSI at the transmitter (CSIT). In p2p multi-antenna systems, CSIT is used to design transmission schemes such as beamforming, spatial power allocation among the antennas and transmit antenna selection, that provide gains in signal to noise ratio (SNR) and diversity. In contrast to a p2p multi-antenna system where the RX antennas can jointly process the received signal and remove the inter-stream interference when multiple parallel data streams are transmitted, the distributed nature of the RXs makes the joint processing impossible in the multi-user system. If the RXs have limited or single antennas, this leaves the TX with the task of designing precoding filters that removes the inter-user interference. In this scenario, CSIT plays an even more crucial role as it allows the TX to serve multiple RXs in parallel. Accurate CSIT plays an important role in achieving the throughput gains with the above mentioned transmission strategies. In practice, CSIT is obtained by a limited rate feedback link in a two-way communication systems where the two links operate in different frequency bands, or by estimating the known pilot symbols sent by the RX when the two links operating on the same frequency band in time-sharing fashion. However, in both systems, acquisition of CSIT creates overhead in the communication resources. The works in [38], [39], have considered training and feedback overhead optimization in a p2p multi-antenna channel where feedback is used in designing beamforming vectors at the TX. An extension to the multi-user scenario where interference canceling precoders are used is analyzed in [40]. A common aspect all these works is that the overhead is modeled in terms of the bandwidth i.e., no of channel uses for sending feedback or number of pilot symbols for training, assuming the RXs have constant power supply. In this thesis, we model the overhead not only in terms of the bandwidth but also in the energy consumed as in our case the terminals depend on limited time-varying harvested energy.

EH Transmitter and Receiver

In this section, we consider the general case where both the TX and the RX harvest energy. Note that if the TX is silent in k-th interval, i.e., pt k = 0, there is no incentive for the RX to send feedback in this interval. Therefore, without loss of optimality we only consider EH profiles where et 1 > 0. Otherwise, if there is an EH profile such that et k = 0; k 2 [1 : m􀀀1], then pt k = 0; k 2 [1 : m􀀀1] due to the constraints in (2.7c). In these intervals the RX simply accumulates the harvested energy, and without loss of optimality we can have a new EH profile with ~et 1 = eti +m􀀀1; 8i 2 [1 : K 􀀀 m + 1], and ~er 1 = Pm k=1 er k and ~eri = eri +m􀀀1; 8i 2 [2 : K 􀀀m+1] for further analysis. We start with the feedback model where UL and DL channel are not identical.

READ THE CONCEPT OF FAITH IN THE GOSPEL OF MARK

Energy Harvesting Model

A time slotted system with K unit duration TSs is considered. At the beginning of the k-th TS, k 2 [1 : K], new energy packets of sizes e1;k and e2;k units arrive at node 1 and 2, respectively. At each node the harvested energy is stored in an infinite size battery and it is used only for communication purposes, i.e., the energy consumed in sampling, compression, etc., is ignored here, and can be considered as an extension. Intially the battery doesnt have any stored energy

Table of contents :

Abstract
Acknowledgements
Contents
List of Figures
Acronyms
Notations
1 Introduction
1.1 Motivation
1.2 Mathematical Model for EH Nodes
1.3 State-of-the-art and Literature Survey
1.3.1 Information Theoretic Studies
1.3.2 Resource Allocation Problems
1.3.3 Multi-antenna Systems with Imperfect CSI
1.4 Outline of the Dissertation
2 Optimization of p2p EH MISO Communication Channels
2.1 Introduction
2.2 Preliminaries
2.3 System model
2.3.1 Energy Harvesting Model
2.3.2 Communication System Model
2.3.3 Feedback Models
2.3.4 Optimization Problem
2.4 EH Receiver
2.4.1 Non-reciprocal channels
2.4.2 Reciprocal channels
2.5 EH Transmitter and Receiver
2.5.1 Non-reciprocal channels
2.5.2 Reciprocal channels
2.6 Numerical Results
2.7 Conclusion
2.8 Appendix
2.8.1 Proof of Lemma 1
2.8.2 Proof of Proposition 1
2.8.3 Proof of Proposition 4
2.8.4 Proof of Proposition 5
2.8.5 Proof of Lemma 4
2.8.6 Proof of Lemma 5
2.8.7 Proof of Proposition 6
3 Training Optimization in TDD MISO Broadcast Channels
3.1 Introduction
3.2 System model
3.2.1 Energy Harvesting Model
3.2.2 Communication System Model
3.2.3 Channel Estimation and Transmission
3.2.4 User Activity
3.2.5 Performance Metric
3.3 Throughput maximization
3.3.1 Approximation
3.4 Greedy user activation
3.5 Numerical Results
3.6 Conclusion
4 Distributed Compression and Transmission with Energy Harvesting Sensors
4.1 Introduction
4.2 System Model
4.2.1 Energy Harvesting Model
4.2.2 Sensing and Communication Model
4.2.3 Problem Formulation
4.3 Characterizing the Pareto boundary of D?
4.3.1 Source coding with a helper node (1 = 0 or 2 = 0)
4.3.2 Weighted sum distortion (1 > 0; 2 > 0)
4.4 Numerical Results
4.5 Conclusion
4.6 Appendix
4.6.1 Proof of Proposition
5 Harvesting and Compression of Information from an Ambient Energy Source
5.1 Introduction
5.2 System model
5.3 Practical Architectures
5.3.1 Energy splitting
5.3.2 Time splitting
5.3.3 Time and energy splitting
5.4 Performance comparison
5.5 QoS with conventional power supply
5.5.1 System model
5.5.2 Numerical Results
5.6 Conclusion
6 Conclusion
7 Résumé
7.1 Résumé [Français]
Bibliography