arXiv subjects

Richard H. Middleton

Publications and source records attributed to Richard H. Middleton.

15 recordsLinked to original sources

arXiv2022

Joint Resource Allocation to Minimize Execution Time of Federated Learning in Cell-Free Massive MIMO

Due to its communication efficiency and privacy-preserving capability, federated learning (FL) has emerged as a promising framework for machine learning in 5G-and-beyond wireless networks. Of great interest is the design and optimization of new wireless network structures that support the stable and fast operation of FL. Cell-free massive multiple-input multiple-output (CFmMIMO) turns out to be a suitable candidate, which allows each communication round in the iterative FL process to be stably executed within a large-scale coherence time. Aiming to reduce the total execution time of the FL process in CFmMIMO, this paper proposes choosing only a subset of available users to participate in FL. An optimal selection of users with favorable link conditions would minimize the execution time of each communication round, while limiting the total number of communication rounds required. Toward this end, we formulate a joint optimization problem of user selection, transmit power, and processing frequency, subject to a predefined minimum number of participating users to guarantee the quality of learning. We then develop a new algorithm that is proven to converge to the neighbourhood of the stationary points of the formulated problem. Numerical results confirm that our proposed approach significantly reduces the FL total execution time over baseline schemes. The time reduction is more pronounced when the density of access point deployments is moderately low.

cs.IT↗

arXiv2021

Discrete-Time Consensus Networks: Scalability, Grounding and Countermeasures

We investigate the disruption of discrete-time consensus problems via grounding. Loosely speaking, grounding a network occurs if the state of one agent no longer responds to inputs from other agents and/or changes its dynamics. Then, the agent becomes a leader or a so-called stubborn agent. The disruption of the agent can be caused by internal faults, safety protocols or due to an external malicious attack. In this paper we investigate how grounding affects expander graph families that usually exhibit good scaling properties with increasing network size. It is shown that the algebraic connectivity and eigenratio of the network decrease due to the grounding causing the performance and scalability of the network to deteriorate, even to the point of losing consensusability. We then present possible countermeasures to such disruptions and discuss their practicality and limitations. In particular, for a specific countermeasure of deliberately grounding additional nodes, we investigate extensively how to select additional nodes to ground and how many nodes we need to ground to recover the consensus performance. Our findings are supported by a wide range of numerical simulations.

eess.SY↗

arXiv2020

Cell-Free Massive MIMO for Wireless Federated Learning

This paper proposes a novel scheme for cell-free massive multiple-input multiple-output (CFmMIMO) networks to support any federated learning (FL) framework. This scheme allows each instead of all the iterations of the FL framework to happen in a large-scale coherence time to guarantee a stable operation of an FL process. To show how to optimize the FL performance using this proposed scheme, we consider an existing FL framework as an example and target FL training time minimization for this framework. An optimization problem is then formulated to jointly optimize the local accuracy, transmit power, data rate, and users' processing frequency. This mixed-timescale stochastic nonconvex problem captures the complex interactions among the training time, and transmission and computation of training updates of one FL process. By employing the online successive convex approximation approach, we develop a new algorithm to solve the formulated problem with proven convergence to the neighbourhood of its stationary points. Our numerical results confirm that the presented joint design reduces the training time by up to $55\%$ over baseline approaches. They also show that CFmMIMO here requires the lowest training time for FL processes compared with cell-free time-division multiple access massive MIMO and collocated massive MIMO.

eess.SP↗

arXiv2020

Scalability and Fragility in Bounded-Degree Consensus Networks

We investigate the performance of linear consensus algorithms subject to a scaling of the underlying network size. Specifically, we model networked systems with $n^{\text{th}}$ order integrator dynamics over families of undirected, weighted graphs with bounded nodal degrees. In such networks, the algebraic connectivity affects convergence rates, sensitivity, and, for high-order consensus ($n \ge 3$), stability properties. This connectivity scales unfavorably in network size, except in expander families, where consensus performs well regardless of network size. We show, however, that consensus over expander families is fragile to a grounding of the network (resulting in leader-follower consensus). We show that grounding may deteriorate system performance by orders of magnitude in large networks, or cause instability in high-order consensus. Our results, which we illustrate through simulations, also point to a fundamental limitation to the scalability of consensus networks with leaders, which does not apply to leaderless networks.

math.OC↗

arXiv2020

Analysis of Attack via Grounding and Countermeasures in Discrete-Time Consensus Networks

We investigate the disruption of discrete-time consensus problems via grounding. Loosely speaking, grounding a network occurs if the state of one agent no longer responds to inputs from other agents and/or changes its dynamics. Then, the agent becomes a leader or a so called stubborn agent. The disruption of the agent can be caused by internal faults, safety protocols or externally due to a malicious attack. In this paper we investigate how the grounding affects the eigenratio of expander graph families that usually exhibit good scaling properties with increasing network size. It is shown that the algebraic connectivity and eigenratio of the network will decrease due to the grounding causing the performance and scalability of the network to deteriorate, even to the point of losing consensusability. We then present countermeasures to such interruptions both in a passive and active manner. Our findings are supported by numerical simulations given within the paper.

eess.SY↗

arXiv2018

A port-Hamiltonian approach to the control of nonholonomic systems

In this paper a method of controlling nonholonomic systems within the port-Hamiltonian (pH) framework is presented. It is well known that nonholonomic systems can be represented as pH systems without Lagrange multipliers by considering a reduced momentum space. Here, we revisit the modelling of these systems for the purpose of identifying the role that physical damping plays. Using this representation, a geometric structure generalising the well known chained form is identified as \textit{chained structure}. A discontinuous control law is then proposed for pH systems with chained structure such that the configuration of the system asymptotically approaches the origin. The proposed control law is robust against the damping and inertial of the open-loop system. The results are then demonstrated numerically on a car-like vehicle.

eess.SY↗

arXiv2018

Discontinuous energy shaping control of the Chaplygin sleigh

In this paper we present an energy shaping control law for set-point regulation of the Chaplygin sleigh. It is well known that nonholonomic mechanical systems cannot be asymptotically stabilised using smooth control laws as they do no satisfy Brockett's necessary condition for smooth stabilisation. Here, we propose a discontinuous control law that can be seen as a potential energy shaping and damping injection controller. The proposed controller is shown to be robust against the parameters of both the inertia matrix and the damping structure of the open-loop system.

eess.SY↗

arXiv2018

Robust integral action of port-Hamiltonian systems

Interconnection and damping assignment, passivity-based control (IDA-PBC) has proven to be a successful control technique for the stabilisation of many nonlinear systems. In this paper, we propose a method to robustify a system which has been stabilised using IDA-PBC with respect to constant, matched disturbances via the addition of integral action. The proposed controller extends previous work on the topic by being robust against the damping of the system, a quantity which may not be known in many applications.

eess.SY↗

arXiv2017

New results on disturbance rejection for energy-shaping controlled port-Hamiltonian systems

In this paper we present a method to robustify energy-shaping controllers for port-Hamiltonian (pH) systems by adding an integral action that rejects unknown additive disturbances. The proposed controller preserves the pH structure and, by adding to the new energy function a suitable cross term between the plant and the controller coordinates, it avoids the unnatural coordinate transformation used in the past. This paper extends our previous work by relaxing the requirement that the systems Hamiltonian is strictly convex and separable, which allows the controller to be applied to a large class of mechanical systems, including underactuated systems with non-constant mass matrix. Furthermore, it is shown that the proposed integral action control is robust against unknown damping in the case of fully-actuated systems.

eess.SY↗

arXiv2017

Matched disturbance rejection for energy-shaping controlled underactuated mechanical systems

In this paper, we present a method of applying integral action to enhance the robustness of energy shaping controllers for underactuated mechanical systems with matched disturbances. Previous works on this problem have required a number of technical assumptions to be satisfied, restricting the class of systems for which the proposed solution applies. The design proposed in this paper relaxes some of these technical assumptions.

eess.SY↗

arXiv2017

Joint Optimization of User Association, Data Delivery Rate and Precoding for Cache-Enabled F-RANs

This paper considers the downlink of a cache-enabled fog radio access network (F-RAN) with limited fronthaul capacity, where user association (UA), data delivery rate (DDR) and signal precoding are jointly optimized. We formulate a mixed-integer nonlinear programming problem in which the weighted difference of network throughput and total power consumption is maximized, subject to the predefined DDR requirements and the maximum transmit power at each eRRH. To address this challenging problem, we first apply the l0-norm approximation and l1-norm minimization techniques to deal with the UA. After this key step, we arrive at an approximated problem that only involves the joint optimization of DDR and precoding. By using the alternating descent method, we further decompose this problem into a convex subproblem for DDR allocation and a nonconvex subproblem for precoding design. While the former is globally solved by the interior-point method, the latter is solved by a specifically tailored successive convex quadratic programming method. Finally, we propose an iterative algorithm for the original joint optimization that is guaranteed to converge. Importantly, each iteration of the developed algorithm only involves solving simple convex problems. Numerical examples demonstrate that the proposed design significantly improves both throughput and power performances, especially in practical F-RANs with limited fronthaul capacity. Compared to the sole precoder design for a given cache placement, our joint design is shown to improve the throughput by 50% while saving at least half of the total power consumption in the considered examples.

cs.IT↗

arXiv2017

Integral control of port-Hamiltonian systems: non-passive outputs without coordinate transformation

In this paper we present a method for the addition of integral action to non-passive outputs of a class of port-Hamiltonian systems. The proposed integral controller is a dynamic extension, constructed from the open loop system, such that the closed loop preserves the port-Hamiltonian form. It is shown that the controller is able to reject the effects of both matched and unmatched disturbances, preserving the regulation of the non-passive outputs. Previous solutions to this problem have relied on a change of coordinates whereas the presented solution is developed using the original state vector and, therefore, retains its physical interpretation. In addition, the resulting closed loop dynamics have a natural interpretation as a Control by Interconnection scheme.

eess.SY↗

arXiv2013

Biological mechanism and identifiability of a class of stationary conductance model for Voltage-gated Ion channels

The physiology of voltage gated ion channels is complex and insights into their gating mechanism is incomplete. Their function is best represented by Markov models with relatively large number of distinct states that are connected by thermodynamically feasible transitions. On the other hand, popular models such as the one of Hodgkin and Huxley have empirical assumptions that are generally unrealistic. Experimental protocols often dictate the number of states in proposed Markov models, thus creating disagreements between various observations on the same channel. Here we aim to propose a limit to the minimum number of states required to model ion channels by employing a paradigm to define stationary conductance in a class of ion-channels. A simple expression is generated using concepts in elementary thermodynamics applied to protein conformational transitions. Further, it matches well many published channel current-voltage characteristics and parameters of the model are found to be identifiable and easily determined from usual experimental protocols.

q-bio.SC↗

arXiv2013

Large-signal stability conditions for semi-quasi-Z-source inverters: switched and averaged models

The recently introduced semi-quasi-Z-source in- verter can be interpreted as a DC-DC converter whose input- output voltage gain may take any value between minus infinity and 1 depending on the applied duty cycle. In order to generate a sinusoidal voltage waveform at the output of this converter, a time-varying duty cycle needs to be applied. Application of a time-varying duty cycle that produces large-signal behavior requires careful consideration of stability issues. This paper provides stability results for both the large-signal averaged and the switched models of the semi-quasi-Z-source inverter operating in continuous conduction mode. We show that if the load is linear and purely resistive then the boundedness and ultimate boundedness of the state trajectories is guaranteed provided some reasonable operation conditions are ensured. These conditions amount to keeping the duty cycle away from the extreme values 0 or 1 (averaged and switched models), and limiting the maximum PWM switching period (switched model). The results obtained can be used to give theoretical justification to the inverter operation strategy recently proposed by Cao et al. in [1].

eess.SY↗

arXiv2011

Towards Modeling HIV Long Term Behavior

The precise mechanism that causes HIV infection to progress to AIDS is still unknown. This paper presents a mathematical model which is able to predict the entire trajectory of the HIV/AIDS dynamics, then a possible explanation for this progression is examined. A dynamical analysis of this model reveals a set of parameters which may produce two real equilibria in the model. One equilibrium is stable and represents those individuals who have been living with HIV for at least 7 to 9 years, and do not develop AIDS. The other one is unstable and represents those patients who developed AIDS in an average period of 10 years. However, further work is needed since the proposed model is sensitive to parameter variations.

q-bio.CB↗