More Like this

1. "Distribution Results for a Multi-Rank Version of the Reed-Yu Detector"

   P. Pakrooh1, L. L. (October 2017, Conference record - Asilomar Conference on Signals, Systems, & Computers)

   In this paper we revisit a detector first derived by Reed and Yu [1], generalized by Bliss and Parker [2], and recently studied by Hiltunen, Loubaton, and Chevalier [3], [4]. The problem is to detect a known signal transmitted over an unknown MIMO channel of unknown complex gains and unknown additive noise covariance. The probability distribution of a CFAR detector for this problem was first derived for the SIMO channel in [1]. We generalize this distribution for the case of a MIMO channel, and show that the CFAR detector statistic is distributed as the product of independent scalar beta random variables under the null. Our results, based on the theory of beta distributed random matrices, hold for M symbols transmitted from p transmitters and received at L receivers. The asymptotic results of [3], [4] are based on large random matrix theory, which assumes L and M to be unbounded. 

   more » « less
2. Linear Stochastic Bandits over a Bit-Constrained Channel

   Mitra, Aritra; Hassani, Hamed; Pappas, George (June 2023, Learning for Dynamics and Control)

   One of the primary challenges in large-scale distributed learning stems from stringent communication constraints. While several recent works address this challenge for static optimization problems, sequential decision-making under uncertainty has remained much less explored in this regard. Motivated by this gap, we introduce a new linear stochastic bandit formulation over a bit-constrained channel. Specifically, in our setup, an agent interacting with an environment transmits encoded estimates of an unknown model parameter to a server over a communication channel of finite capacity. The goal of the server is to take actions based on these estimates to minimize cumulative regret. To this end, we develop a novel and general algorithmic framework that hinges on two main components: (i) an adaptive encoding mechanism that exploits statistical concentration bounds, and (ii) a decision-making principle based on confidence sets that account for encoding errors. As our main result, we prove that when the unknown model is d-dimensional, a channel capacity of O(d) bits suffices to achieve order-optimal regret. We also establish that for the simpler unstructured multi-armed bandit problem, 1 bit channel capacity is sufficient for achieving optimal regret bounds. 

   more » « less
3. Linear stochastic bandits over a bit-constrained channel

   Mitra Aritra; Hassani, Hamed; Pappas, George (June 2023, Learning for Dynamics and Control Conference)

   One of the primary challenges in large-scale distributed learning stems from stringent communication constraints. While several recent works address this challenge for static optimization problems, sequential decision-making under uncertainty has remained much less explored in this regard. Motivated by this gap, we introduce a new linear stochastic bandit formulation over a bit-constrained channel. Specifically, in our setup, an agent interacting with an environment transmits encoded estimates of an unknown model parameter to a server over a communication channel of finite capacity. The goal of the server is to take actions based on these estimates to minimize cumulative regret. To this end, we develop a novel and general algorithmic framework that hinges on two main components:(i) an adaptive encoding mechanism that exploits statistical concentration bounds, and (ii) a decision-making principle based on confidence sets that account for encoding errors. As our main result, we prove that when the unknown model is -dimensional, a channel capacity of bits suffices to achieve order-optimal regret. We also establish that for the simpler unstructured multi-armed bandit problem, bit channel capacity is sufficient for achieving optimal regret bounds. 

   more » « less
4. Deep Neural Network Based Media Noise Predictors for Use in High-Density Magnetic Recording Turbo-Detectors

   Sayyafan, Amirhossen; Belzer, Benjamin J; Sivakjumar, Krishnamoorthy; Shen, Jinlu; Chan, Kheong Sann; James, Ashish (July 2019, 2019 Magnetic Recording Conference (TMRC 2019))

   Trellis based detection with pattern dependent noise prediction (PDNP) has become standard practice in the HDD industry. In a typical single-track signal processing scheme, the received samples from the read head are first filtered by a linear equalizer with a 1D partial response (PR). The linear filter output flows into a trellis-based (e.g. BCJR) detector that employs a super-trellis based on the PR mask ISI channel and a 1D pattern dependent noise prediction (1D PDNP) algorithm. The effective channel model has a media noise term which models signal dependent noise due to, e.g., magnetic grains intersected by bit boundaries. The media noise can influence two or more bit readback values. The trellis detector sends soft estimates of the coded bits to a channel decoder, which outputs estimates of the user information bits. There are two problems with traditional PDNP. First, when the number of tracks Nt simultaneously processed is greater than one, e.g. in two-dimensional magnetic recording (TDMR), the number of trellis states can become impractically large; this is the state explosion problem. Second, the relatively simple autoregressive noise model and linear prediction used in PDNP is somewhat restrictive and may not accurately represent the media noise, especially at high storage densities; this is the modeling problem. To address the state explosion problem, we separate the ISI detection and media noise prediction into two separate detectors and use the turbo-principle to exchange information between them, thus avoiding use of a super-trellis. To address the modeling problem, we design and train deep neural network (DNN) based media noise predictors. As DNN models are much more general than autoregressive models, they give a more accurate model of magnetic media noise than PDNP; this more accurate modeling results in reduced detector BERs compared to PDNP. 

   more » « less
5. Low-Complexity, Low-Regret Link Rate Selection in Rapidly Time-Varying Wireless Channels

   H. Gupta, A. Eryilmaz (April 2018, INFOCOMP ... the ... International Conference on Advanced Communications and Computation)

   We consider the problem of transmitting at the optimal rate over a rapidly-varying wireless channel with unknown statistics when the feedback about channel quality is very limited. One motivation for this problem is that, in emerging wireless networks, the use of mmWave bands means that the channel quality can fluctuate rapidly and thus, one cannot rely on full channel-state feedback to make transmission rate decisions. Inspired by related problems in the context of multi-armed bandits, we consider a well-known algorithm called Thompson sampling to address this problem. However, unlike the traditional multi-armed bandit problem, a direct application of Thompson sampling results in a computational and storage complexity that grows exponentially with time. Therefore, we propose an algorithm called Modified Thompson sampling (MTS), whose computational and storage complexity is simply linear in the number of channel states and which achieves at most logarithmic regret as a function of time when compared to an optimal algorithm which knows the probability distribution of the channel states. 

   more » « less