Between Dec 19, 2024 and Jan 2, 2025, datasets can be submitted to DRUM but will not be processed until after the break. Staff will not be available to answer email during this period, and will not be able to provide DOIs until after Jan 2. If you are in need of a DOI during this period, consider Dryad or OpenICPSR. Submission responses to the UDC may also be delayed during this time.

Browsing by Subject "Binary CSIT"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    High Performance Adaptive Transmit Beamforming forWireless Networks using Binary CSIT
    (2015-05) Gopalakrishnan, Balasubramanian
    Transmit beamforming is a characteristic feature of the modern wireless communication standards like 4G-LTE and 802.11 ac because of the increasing demand for higher data rates and better Quality of Service at the user-end. Transmit beamforming uses multiple transmit antennas and channel state information (CSI) at the transmitter (Tx) to steer the radiated power towards the intended receiver (Rx) while limiting the leakage caused in other directions. In the absence of channel reciprocity, this channel information is acquired at the transmitter by channel estimation at the intended Rx and subsequent feedback of the quantized channel information back to the Tx. This conventional training method requires a complex Rx design and high communication overhead, which could be a burden when the receivers operate on battery power and have limited computational resources and restricted communication capabilities. Obtaining CSI for limiting interference caused due to leakage, can be much more challenging especially when the Rx affected by the spatial leakage interference (side lobes) is not cooperating with the Tx (as in secondary transmit beamforming in underlay cognitive radio networks, for achieving a high Quality of Service at the secondary Rx while limiting the interference to the primary Rx). This thesis proposes various algorithms that enable the Tx, which has no initial CSI, learn to beamform on-the-fly and asymptotically attain the performance achievable using perfect CSI at the Tx, using 1-bit direct or implicit periodic feedbacks from the receivers of interest. The receivers are assumed to have limited computational capability. This thesis starts by considering long-term transmit beamforming for point-to-point Multiple-Input Single-Output (MISO) links and proposes an online beamforming and learning algorithm using the analytic center cutting plane method which is shown to asymptotically attain optimal performance. A robust maximum-likelihood formulation is next developed to combat feedback errors and correlation drift. The setup is then extended to an underlay cognitive radio network for designing secondary transmit beamforming vectors that maximize the average Signal-to-Noise Ratio (SNR) at the secondary Rx while limiting the interference to the primary Rx, using direct binary channel quality indicator feedback bits from the secondary Rx and indirect ACK-NACK feedback from the primary Rx. When the primary interference threshold is known at the secondary Tx, it is analytically shown that the proposed algorithm converges to maximum average SNR at the secondary Rx achieved using perfect CSI at the Tx. Subsequently, the thesis considers max-min fair transmit beamforming for single group multicast networks (which is NP-hard in general) and introduces a new class of adaptive beamforming algorithms that features guaranteed convergence and state-of-the-art performance at low complexity, when perfect CSI is available at the Tx. Convergence to a Karush-Kuhn-Tucker (KKT) point of a related proportionally fair beamforming is established. Simulations show that the proposed approach outperforms the prior state-of-art in terms of multicast rate, at considerably lower complexity. When there is no initial CSIT, an extension of the online algorithm developed for point to point MISO links is proposed for designing the beamforming vector to maximize the minimum SNR among the users, using only periodic binary SNR feedback from each Rx. The design methodology for the multicast beamforming problem is finally extended in a novel fashion to obtain feasible solutions for non-convex Quadratically Constrained Quadratic Programs (QCQP) with two-sided constraints when the associated matrices are positive semi-definite. In this context, the proposed algorithm starts with a infeasible solution which is iteratively updated using a gradient of the log-barrier function of the non-convex constraints followed by projection onto the intersection of the set of convex constraints and a refining step using successive linear approximation. Convergence of the algorithm is established using the Descent lemma and simulations show that the algorithm obtains feasible solutions with a high probability at a much lower complexity compared to the state-of-the-art.