An actor-critic algorithm with function approximation for discounted cost constrained Markov decision processes

We develop in this article the first actor-critic reinforcement learning algorithm with function approximation for a problem of control under multiple inequality constraints. We consider the infinite horizon discounted cost framework in which both the objective and the constraint functions are suitable expected policy-dependent discounted sums of certain sample path functions. We apply the Lagrange multiplier method to handle the inequality constraints. Our algorithm makes use of multi-timescale stochastic approximation and incorporates a temporal difference (TD) critic and an actor that makes a gradient search in the space of policy parameters using efficient simultaneous perturbation stochastic approximation (SPSA) gradient estimates. We prove the asymptotic almost sure convergence of our algorithm to a locally optimal policy. (C) 2010 Elsevier B.V. All rights reserved.

Study Of Acoustic Source Localization Algorithm For Planar Arrays

The source localization algorithms in the earlier works, mostly used non-planar arrays. If we consider scenarios like human-computer communication, or human-television communication where the microphones need to be placed on the computer monitor or television front panel, i.e we need to use the planar arrays. The algorithm proposed in 1], is a Linear Closed Form source localization algorithm (LCF algorithm) which is based on Time Difference of Arrivals (TDOAs) that are obtained from the data collected using the microphones. It assumes non-planar arrays. The LCF algorithm is applied to planar arrays in the current work. The relationship between the error in the source location estimate and the perturbation in the TDOAs is derived using first order perturbation analysis and validated using simulations. If the TDOAs are erroneous, both the coefficient matrix and the data matrix used for obtaining source location will be perturbed. So, the Total least squares solution for source localization is proposed in the current work. The sensitivity analysis of the source localization algorithm for planar arrays and non-planar arrays is done by introducing perturbation in the TDOAs and the microphone locations. It is shown that the error in the source location estimate is less when we use planar array instead of the particular non-planar array considered for same perturbation in the TDOAs or microphone location. The location of the reference microphone is proved to be important for getting an accurate source location estimate if we are using the LCF algorithm.

Past Queue Length Based Low-Overhead Link Scheduling in Multi-beam Wireless Mesh Networks

Wireless mesh networks with multi-beam capability at each node through the use of multi-antenna beamforming are becoming practical and attracting increased research attention. Increased capacity due to spatial reuse and increased transmission range are potential benefits in using multiple directional beams in each node. In this paper, we are interested in low-complexity scheduling algorithms in such multi-beam wireless networks. In particular, we present a scheduling algorithm based on queue length information of the past slots in multi-beam networks, and prove its stability. We present a distributed implementation of this proposed algorithm. Numerical results show that significant improvement in delay performance is achieved using the proposed multi-beam scheduling compared to omni-beam scheduling. In addition, the proposed algorithm is shown to achieve a significant reduction in the signaling overhead compared to a current slot queue length approach.

Explicit and Optimal Exact-Regenerating Codes for the Minimum-Bandwidth Point in Distributed Storage

In the distributed storage setting that we consider, data is stored across n nodes in the network such that the data can be recovered by connecting to any subset of k nodes. Additionally, one can repair a failed node by connecting to any d nodes while downloading beta units of data from each. Dimakis et al. show that the repair bandwidth d beta can be considerably reduced if each node stores slightly more than the minimum required and characterize the tradeoff between the amount of storage per node and the repair bandwidth. In the exact regeneration variation, unlike the functional regeneration, the replacement for a failed node is required to store data identical to that in the failed node. This greatly reduces the complexity of system maintenance. The main result of this paper is an explicit construction of codes for all values of the system parameters at one of the two most important and extreme points of the tradeoff - the Minimum Bandwidth Regenerating point, which performs optimal exact regeneration of any failed node. A second result is a non-existence proof showing that with one possible exception, no other point on the tradeoff can be achieved for exact regeneration.

A Flexible Class of Regenerating Codes for Distributed Storage

In the distributed storage setting introduced by Dimakis et al., B units of data are stored across n nodes in the network in such a way that the data can be recovered by connecting to any k nodes. Additionally one can repair a failed node by connecting to any d nodes while downloading at most beta units of data from each node. In this paper, we introduce a flexible framework in which the data can be recovered by connecting to any number of nodes as long as the total amount of data downloaded is at least B. Similarly, regeneration of a failed node is possible if the new node connects to the network using links whose individual capacity is bounded above by beta(max) and whose sum capacity equals or exceeds a predetermined parameter gamma. In this flexible setting, we obtain the cut-set lower bound on the repair bandwidth along with a constructive proof for the existence of codes meeting this bound for all values of the parameters. An explicit code construction is provided which is optimal in certain parameter regimes.

Generating synchronizable test sequences based on finite-statemachine with distributed ports

In the area of testing communication systems, the interfaces between systems to be tested and their testers have great impact on test generation and fault detectability. Several types of such interfaces have been standardized by the International Standardization Organization (ISO). A general distributed test architecture, containing distributed interfaces, has been presented in the literature for testing distributed systems based on the Open Distributing Processing (ODP) Basic Reference Model (BRM), which is a generalized version of ISO distributed test architecture. We study in this paper the issue of test selection with respect to such an test architecture. In particular, we consider communication systems that can be modeled by finite state machines with several distributed interfaces, called ports. A test generation method is developed for generating test sequences for such finite state machines, which is based on the idea of synchronizable test sequences. Starting from the initial effort by Sarikaya, a certain amount of work has been done for generating test sequences for finite state machines with respect to the ISO distributed test architecture, all based on the idea of modifying existing test generation methods to generate synchronizable test sequences. However, none studies the fault coverage provided by their methods. We investigate the issue of fault coverage and point out a fact that the methods given in the literature for the distributed test architecture cannot ensure the same fault coverage as the corresponding original testing methods. We also study the limitation of fault detectability in the distributed test architecture.

Multi-Agent Rendezvous Algorithm with Rectilinear Decision Domain

The aim of this paper is to develop a computationally efficient decentralized rendezvous algorithm for a group of autonomous agents. The algorithm generalizes the notion of sensor domain and decision domain of agents to enable implementation of simple computational algorithms. Specifically, the algorithm proposed in this paper uses a rectilinear decision domain (RDD) as against the circular decision domain assumed in earlier work. Because of this, the computational complexity of the algorithm reduces considerably and, when compared to the standard Ando's algorithm available in the literature, the RDD algorithm shows very significant improvement in convergence time performance. Analytical results to prove convergence and supporting simulation results are presented in the paper.

Layered Tabu Search Algorithm for Large-MIMO Detection and a Lower Bound on ML Performance

In this paper, we are concerned with low-complexity detection in large multiple-input multiple-output (MIMO) systems with tens of transmit/receive antennas. Our new contributions in this paper are two-fold. First, we propose a low-complexity algorithm for large-MIMO detection based on a layered low-complexity local neighborhood search. Second, we obtain a lower bound on the maximum-likelihood (ML) bit error performance using the local neighborhood search. The advantages of the proposed ML lower bound are i) it is easily obtained for MIMO systems with large number of antennas because of the inherent low complexity of the search algorithm, ii) it is tight at moderate-to-high SNRs, and iii) it can be tightened at low SNRs by increasing the number of symbols in the neighborhood definition. Interestingly, the proposed detection algorithm based on the layered local search achieves bit error performances which are quite close to this lower bound for large number of antennas and higher-order QAM. For e. g., in a 32 x 32 V-BLAST MIMO system, the proposed detection algorithm performs close to within 1.7 dB of the proposed ML lower bound at 10(-3) BER for 16-QAM (128 bps/Hz), and close to within 4.5 dB of the bound for 64-QAM (192 bps/Hz).

Managing interprocessor delays in distributed recursive algorithms

For a class of distributed recursive algorithms, it is shown that a stochastic approximation-like tapering stepsize routine suppresses the effects of interprocessor delays.

Performance of a pipelined ring algorithm for Fast Fourier Transform on transputer arrays

In this paper a pipelined ring algorithm is presented for efficient computation of one and two dimensional Fast Fourier Transform (FFT) on a message passing multiprocessor. The algorithm has been implemented on a transputer based system and experiments reveal that the algorithm is very efficient. A model for analysing the performance of the algorithm is developed from its computation-communication characteristics. Expressions for execution time, speedup and efficiency are obtained and these expressions are validated with experimental results obtained on a four transputer system. The analytical model is then used to estimate the performance of the algorithm for different number of processors, and for different sizes of the input data.

Additive Spanners and (alpha, beta)-Spanners

An (alpha, beta)-spanner of an unweighted graph G is a subgraph H that distorts distances in G up to a multiplicative factor of a and an additive term beta. It is well known that any graph contains a (multiplicative) (2k - 1, 0)-spanner of size O(n(1+1/k)) and an (additive) (1, 2)-spanner of size O(n(3/2)). However no other additive spanners are known to exist. In this article we develop a couple of new techniques for constructing (alpha, beta)-spanners. Our first result is an additive (1, 6)-spanner of size O(n(4/3)). The construction algorithm can be understood as an economical agent that assigns costs and values to paths in the graph, purchasing affordable paths and ignoring expensive ones, which are intuitively well approximated by paths already purchased. We show that this path buying algorithm can be parameterized in different ways to yield other sparseness-distortion tradeoffs. Our second result addresses the problem of which (alpha, beta)-spanners can be computed efficiently, ideally in linear time. We show that, for any k, a (k, k - 1)-spanner with size O(kn(1+1/k)) can be found in linear time, and, further, that in a distributed network the algorithm terminates in a constant number of rounds. Previous spanner constructions with similar performance had roughly twice the multiplicative distortion.

Stabilizing trench's algorithm to invert symmetric toeplitz matrices

Presented here is a stable algorithm that uses Zohar's formulation of Trench's algorithm and computes the inverse of a symmetric Toeplitz matrix including those with vanishing or nearvanishing leading minors. The algorithm is based on a diagonal modification of the matrix, and exploits symmetry and persymmetry properties of the inverse matrix.

A Flat Concurrent Prolog compiler for PARAM

We describe a compiler for the Flat Concurrent Prolog language on a message passing multiprocessor architecture. This compiler permits symbolic and declarative programming in the syntax of Guarded Horn Rules, The implementation has been verified and tested on the 64-node PARAM parallel computer developed by C-DAC (Centre for the Development of Advanced Computing, India), Flat Concurrent Prolog (FCP) is a logic programming language designed for concurrent programming and parallel execution, It is a process oriented language, which embodies dataflow synchronization and guarded-command as its basic control mechanisms. An identical algorithm is executed on every processor in the network, We assume regular network topologies like mesh, ring, etc, Each node has a local memory, The algorithm comprises of two important parts: reduction and communication, The most difficult task is to integrate the solutions of problems that arise in the implementation in a coherent and efficient manner. We have tested the efficacy of the compiler on various benchmark problems of the ICOT project that have been reported in the recent book by Evan Tick, These problems include Quicksort, 8-queens, and Prime Number Generation, The results of the preliminary tests are favourable, We are currently examining issues like indexing and load balancing to further optimize our compiler.

Optimal management of beaver population using a reduced-order distributed parameter model and single network adaptive critics

Beavers are often found to be in conflict with human interests by creating nuisances like building dams on flowing water (leading to flooding), blocking irrigation canals, cutting down timbers, etc. At the same time they contribute to raising water tables, increased vegetation, etc. Consequently, maintaining an optimal beaver population is beneficial. Because of their diffusion externality (due to migratory nature), strategies based on lumped parameter models are often ineffective. Using a distributed parameter model for beaver population that accounts for their spatial and temporal behavior, an optimal control (trapping) strategy is presented in this paper that leads to a desired distribution of the animal density in a region in the long run. The optimal control solution presented, imbeds the solution for a large number of initial conditions (i.e., it has a feedback form), which is otherwise nontrivial to obtain. The solution obtained can be used in real-time by a nonexpert in control theory since it involves only using the neural networks trained offline. Proper orthogonal decomposition-based basis function design followed by their use in a Galerkin projection has been incorporated in the solution process as a model reduction technique. Optimal solutions are obtained through a "single network adaptive critic" (SNAC) neural-network architecture.

Reduced Order Suboptimal Control Design for a Class of Nonlinear Distributed Parameter Systems Using POD and /spl thetas/-D Techniques

A new computational tool is presented in this paper for suboptimal control design of a class of nonlinear distributed parameter systems. First proper orthogonal decomposition based problem-oriented basis functions are designed, which are then used in a Galerkin projection to come up with a low-order lumped parameter approximation. Next, a suboptimal controller is designed using the emerging /spl thetas/-D technique for lumped parameter systems. This time domain sub-optimal control solution is then mapped back to the distributed domain using the same basis functions, which essentially leads to a closed form solution for the controller in a state feedback form. Numerical results for a real-life nonlinear temperature control problem indicate that the proposed method holds promise as a good suboptimal control design technique for distributed parameter systems.