This paper studies the transmission of a stream of data packets through a multi-hop multi-link communication network. The source generate sequential source packets and transmits them through a network to a destination. Each intermediary node is allowed to perform any processing operation in order to successfully forward the message to the destination. Each link connecting two nodes is subject to an adversarial packet erasure channel, where the number of erasures is limited. The destination must recover each source packet within a delay deadline, otherwise that packet is considered lost. The particular case when each hop consists of only one link has been studied by Domanovitz et. al in [1]. We extend this work to the multi-link scenario by multiplexing multiple copies of streaming codes in [1] and optimizing the rate using a linear programming framework. We also demonstrate how our framework can be naturally extended when there is a different propagation delay on each link. Numerical results demonstrate that our proposed approach provides significant improvements over baseline schemes.We further demonstrate that our proposed scheme also provides improved audio quality over baseline schemes in experiments involving Gilbert-Elliott channels.

This paper studies packet repetition strategies over erasure channels with memory and a long feedback delay. The problem is initially formulated as a communications problem where a source wishes to transmit one message packet to a destination while minimizing both the delay and the number of transmissions. At each time instant, the sender is provided a delayed acknowledgement feedback about past attempts, and must decide whether to attempt a new transmission or not. This problem is then re-formulated as an episodic reinforcement learning problem, where an agent attempts to learn the optimal transmission policy, provided delayed feedback about past transmission attempts. The agent is helped by a channel estimator, which attempts to capture the channel memory and use that to predict probabilities of erasures in a future window. This channel estimator is also data-driven and learns the channel model without any a priori channel knowledge. The paper presents a lower bound on the achievable trade-off between delay and number of transmissions for any channel modeled as a Markov process. Experimental results show that the combination of the proposed channel estimator and the agent can noticeably outperform naive strategies for channels with memory, and achieves results close to the lower bound.

This paper investigates the performance of streaming codes in low-latency applications over a multi-link three-node relayed network. The source wishes to transmit a sequence of messages to the destination through a relay. Each message must be reconstructed after a fixed decoding delay. The special case with one link connecting each node has been studied by Fong et. al [1], and a multi-hop multi-link setting has been studied by Domanovitz et. al [2]. The topology with three nodes and multiple links is studied in this paper. Each link is subject to a different number of erasures due to different channel conditions. An information-theoretic upper bound is derived, and an achievable scheme is presented. The proposed scheme judiciously allocates rates for each link based on the concept of delay spectrum. The achievable scheme is compared to two baseline schemes and the scheme proposed in [2]. Experimental results show that this scheme achieves higher rates than the other schemes, and can achieve the upper bound even in non-trivial scenarios. The scheme is further extended to handle different propagation delays in each link, something not previously considered in the literature. Simulations over statistical channels show that the proposed scheme can outperform the simpler baseline under practical models.

This paper studies low-latency streaming codes for the multi-hop network. The source transmits a sequence of messages to a destination through a chain of relays, and requires the destination to reconstruct each message by its deadline. We assume that each communication link is subjected to a certain maximum number of packet erasures. The case of a single relay (a three-node network) was considered in Fong et al. (2020). A coding scheme known as symbol-wise decode and forward was proposed. In the present work, we propose an alternative scheme that is different from Fong et al. (2020) and still achieves the same rate as in Fong et al. (2020) for the one hop case as the field-size goes to infinity. Furthermore, our proposed scheme naturally generalizes to the case of multiple-relay nodes yielding new achievable rates for this setting. The main difference with Fong et al. (2020) is that our proposed scheme exploits the ability of the relay nodes to adapt the transmission based on the erasures on the previous link. Hence, we refer to our scheme as “state-dependent” and contrast it with the scheme in Fong et al. (2020) that is state-independent. Our scheme requires the relay nodes to append a header to the transmitted packets, and we show that the size of the header does not depend on the field-size of the code. We also derive an upper bound on the maximal streaming rate achievable over a network with an arbitrary number of relays. We show that this upper bound matches our achievable rate in the special case when the maximal number of erasures on the first link is greater than or equal to the maximal number of erasures on each of the following links, and the field size goes to infinity.

We introduce a novel network-adaptive algorithm that is suitable for alleviating network packet losses for low-latency interactive communications between a source and a destination. Our network-adaptive algorithm estimates in real-time the best parameters of a recently proposed streaming code that uses forward error correction (FEC) to correct both arbitrary and burst losses, which cause a crackling noise and undesirable jitters, respectively in audio. In particular, the destination estimates appropriate coding parameters based on its observed packet loss pattern and sends them back to the source for updating the underlying code. Besides, a new explicit construction of practical low-latency streaming codes that achieve the optimal tradeoff between the capability of correcting arbitrary losses and the capability of correcting burst losses is used. Simulation evaluations based on statistical losses and real-world packet loss traces reveal the following: (i) Our proposed network-adaptive algorithm combined with our optimal streaming codes can achieve significantly higher performance compared to uncoded and non-adaptive FEC schemes over UDP (User Datagram Protocol); (ii) Our explicit streaming codes can significantly outperform traditional MDS (maximum-distance separable) streaming schemes when they are used along with our network-adaptive algorithm. In addition, we study different factors that can affect the performance of our network-adaptive algorithm.

Manifold landmarking is the problem of selecting a subset of discrete locations on a continuous manifold for label assignment, in order to reduce interpolation error of subsequent semi-supervised learning. In this paper, we select landmarks to minimize the condition number (λmax/λmin) of a submatrix of an alignment matrix Φ, which is equivalent to minimizing an interpolation error bound. Specifically, we design an efficient greedy scheme, where at each iteration t + 1 we choose one landmark i (thus deleting the corresponding row and column i of Φt) so that the resulting submatrix Φt+1 has the smallest condition number. Towards fast landmark selection, at iteration t + 1, we first compute the two extreme eignevectors v1 and vN corresponding to λmin and λmax of Φt via known methods like LOBPCG. We show that λmin (λmax) of submatrix Φt+1, from deleting the chosen row-column pair, can be approximated by an upper (lower) bound that is an easily computable function of eigen-pair {v1, λmin} ({vN, λmax}) of Φt. The error bounds of the obtained approximations can be numerically computed during the greedy step. Leveraging these proofs, we minimize a bound of the condition number for submatrix Φt+1 at each greedy step t + 1. Experiments on synthetic and real-world manifold data demonstrate the superiority of our proposed landmarking algorithm compared to several state-of-the-art schemes.

Multi-user MIMO (MU-MIMO) is a technique that improves spectral efficiency by allowing concurrent communication between one access point (AP) and multiple clients. In practice, the expected gain is not always achieved and is sometimes even negative. We experimentally demonstrate that the downlink MU-MIMO performance in a practical network not only depends on the client's channel but is also influenced by factors that are not captured by conventional models, such as client motion and device type. We propose a data-driven algorithm with a low computational complexity that determines whether a client should operate in MU mode and the MU-MIMO group for clients in MU mode. Such a mode and group selection algorithm is based on a sequence of channel state information (CSI), SNR, and client device type. The algorithm can automatically adapt to the motion and characteristics of individual clients. Experimental results using implementation on a commodity 802.11ac AP show that the proposed data-driven mode and group selection algorithm can improve network throughput by up to 35% over existing algorithms based on conventional models. We also show that the proposed data-driven algorithm has limited sensitivity to environmental changes and can be deployed into new environments without retraining.

In this work, we develop DeepWiPHY, a deep learning-based architecture to replace the channel estimation, common phase error (CPE) correction, sampling rate offset (SRO) correction, and equalization modules of IEEE 802.11ax based orthogonal frequency division multiplexing (OFDM) receivers. We first train DeepWiPHY with a synthetic dataset, which is generated using representative indoor channel models and includes typical radio frequency (RF) impairments that are the source of nonlinearity in wireless systems. To further train and evaluate DeepWiPHY with real-world data, we develop a passive sniffing-based data collection testbed composed of Universal Software Radio Peripherals (USRPs) and commercially available IEEE 802.11ax products. The comprehensive evaluation of DeepWiPHY with synthetic and real-world datasets (110 million synthetic OFDM symbols and 14 million real-world OFDM symbols) confirms that, even without fine-tuning the neural network's architecture parameters, DeepWiPHY achieves comparable performance to or outperforms the conventional WLAN receivers, in terms of both bit error rate (BER) and packet error rate (PER), under a wide range of channel models, signal-to-noise (SNR) levels, and modulation schemes.

We study the problem of transmitting a sequence of messages (streaming messages) through a multi-link, multi-hop packet erasure network. Each message must be reconstructed in-order and under a strict delay constraint. Special cases of our setting with a single link on each hop have been studied recently - the case of a single relay-node, is studied in Fong et al [1]; the case of multiple relays, is studied in Domanovitz et al [2]. As our main result, we propose an achievable rate expression that reduces to previously known results when specialized to their respective settings. Our proposed scheme is based on the idea of concatenating single-link codes from [2] in a judicious manner to achieve the required delay constraints. We propose a systematic approach based on convex optimization to maximize the achievable rate in our framework.

Multi-User Multiple-Input and Multiple-Output (MU-MIMO) is a technique that allows concurrent transmissions between one access point (AP) and multiple clients to improve spectral efficiency. In practice, however, the MU-MIMO is sensitive to client mobility and is sometimes even harmful to the performance in networks with moving clients. In this paper, we identify that it is essential to optimize the MU-MIMO performance with moving clients by jointly selecting the sounding period, the number of spatial streams, and client grouping with the consideration of the client density of the network. We develop a data-driven model that estimates client throughput with the consideration of these parameters, as well as an algorithm that jointly determines the parameters for each client with low computational complexity. Using a commodity 802.11ax network, we experimentally demonstrate the significant impact of the key factors on MU-MIMO performance. Based on experimental data, we develop an emulation model to evaluate network performance with different client densities and mobility. Emulation results show that our proposed algorithm outperforms conventional schemes by over 20% in MU-MIMO networks with moving clients.

In semi-supervised graph-based binary classifier learning, a subset of known labels $\hat{x}_i$ are used to infer unknown labels, assuming that the label signal $\x$ is smooth with respect to a similarity graph specified by a Laplacian matrix. When restricting labels $x_i$ to binary values, the problem is NP-hard. While a conventional semi-definite programming (SDP) relaxation can be solved in polynomial time using, for example, the alternating direction method of multipliers (ADMM), the complexity of iteratively projecting a candidate matrix $\M$ onto the positive semi-definite (PSD) cone ($\M \succeq 0$) remains high. In this paper, leveraging a recent linear algebraic theory called Gershgorin disc perfect alignment (GDPA), we propose a fast projection-free method by solving a sequence of linear programs (LP) instead. Specifically, we first recast the SDP relaxation to its SDP dual, where a feasible solution $\H \succeq 0$ can be interpreted as a Laplacian matrix corresponding to a balanced signed graph sans the last node. To achieve graph balance, we split the last node into two that respectively contain the original positive and negative edges, resulting in a new Laplacian $\bar{\H}$. We repose the SDP dual for solution $\bar{\H}$, then replace the PSD cone constraint $\bar{\H} \succeq 0$ with linear constraints derived from GDPA---sufficient conditions to ensure $\bar{\H}$ is PSD---so that the optimization becomes an LP per iteration. Finally, we extract predicted labels from our converged LP solution $\bar{\H}$. Experiments show that our algorithm enjoyed a $40\times$ speedup on average over the next fastest scheme while retaining comparable label prediction performance.

Existing Active Learning (AL) schemes typically address privacy in the narrow sense of furnishing a differentially private classifier. Private data are exposed to both the labeling and learning functions, a limitation that necessarily restricts their applicability to a single entity. In this paper, we propose an AL framework that allows the use of untrusted parties for both labeling and learning, thereby allowing joint use of data from multiple entities without trust relationships. Our method is based on differentially private generative models and an associated novel latent space optimization scheme that is more flexible than the traditional ranking method. Our experiments on three datasets (MNIST, CIFAR10, CelebA) show that our proposed scheme produces better or compara-ble results than state-of-the-art techniques on two different acquisition functions (VAR and BALD).

In this paper, we investigate streaming codes over a three-node relay network. Source node transmits a sequence of message packets to the destination via a relay. Source-to-relay and relay-to-destination links are unreliable and introduce at most N 1 and N 2 packet erasures, respectively. Destination needs to recover each message packet with a strict decoding delay constraint of T time slots. We propose streaming codes under this setting for all feasible parameters $\{N_{1},\ N_{2},\ T\}$. Relay naturally observes erasure patterns occurring in the source-to-relay link. In our code construction, we employ a channel-state-dependent relaying strategy, which rely on these observations. In a recent work, Fong et al. provide streaming codes featuring channel-state-independent relaying strategies, for all feasible parameters $\{N_{1},\ N_{2},\ T\}$. Our schemes offer a strict rate improvement over the schemes proposed by Fong et al., whenever $N_{1}\lt N_{2}$.

Applications where multiple users communicate with a common server and desire low latency are common and increasing. This paper studies a network with two source nodes, one relay node and a destination node, where each source nodes wishes to transmit a sequence of messages, through the relay, to the destination, who is required to decode the messages with a strict delay constraint $T$. The network with a single source node has been studied in \cite{Silas2019}. We start by introducing two important tools: the delay spectrum, which generalizes delay-constrained point-to-point transmission, and concatenation, which, similar to time sharing, allows combinations of different codes in order to achieve a desired regime of operation. Using these tools, we are able to generalize the two schemes previously presented in \cite{Silas2019}, and propose a novel scheme which allows us to achieve optimal rates under a set of well-defined conditions. Such novel scheme is further optimized in order to improve the achievable rates in the scenarios where the conditions for optimality are not met.

Algorithm unfolding creates an interpretable and parsimonious neural network architecture by implementing each iteration of a model-based algorithm as a neural layer. However, unfolding a proximal splitting algorithm with a positive semi-definite (PSD) cone projection operator per iteration is expensive, due to the required full matrix eigen-decomposition. In this paper, leveraging a recent linear algebraic theorem called Gershgorin disc perfect alignment (GDPA), we unroll a projection-free algorithm for semi-definite programming relaxation (SDR) of a binary graph classifier, where the PSD cone constraint is replaced by a set of "tightest possible" linear constraints per iteration. As a result, each iteration only requires computing a linear program (LP) and one extreme eigenvector. Inside the unrolled network, we optimize parameters via stochastic gradient descent (SGD) that determine graph edge weights in two ways: i) a metric matrix that computes feature distances, and ii) a sparse weight matrix computed via local linear embedding (LLE). Experimental results show that our unrolled network outperformed pure model-based graph classifiers, and achieved comparable performance to pure data-driven networks but using far fewer parameters.

Many emerging multimedia streaming applications involve multiple users communicating under strict latency constraints. In this paper we study streaming codes for a network involving two source nodes, one relay node and a destination node. In this paper's setting, each source node transmits a stream of messages, through the relay, to a destination, who is required to decode the messages under a strict delay constraint. For the case of a single source node, a class of streaming codes has been proposed by Fong et al., using the concept of delay-spectrum. The current paper presents a novel framework, which constructs streaming codes for a relayed multi-user setting by sequentially constructing the codes for each link. This requires a characterization of the set of all achievable delay spectra for a given rate, blocklength and number of erasures, beyond the specific choice considered by Fong et al. This characterization is presented in the paper for systematic codes. Using this novel framework, the first proposed scheme involves greedily selecting the rate on the link from relay to destination and using properties of the delay-spectrum to find feasible streaming codes that satisfy the required delay constraints. A closed form expression for the achievable rate region is provided, and conditions for when the proposed scheme is optimal are established by a natural outer bound. The second proposed scheme builds upon this approach, but uses a numerical optimization-based approach to improve the achievable rate region over the first scheme. Experimental results show that the proposed schemes achieve significant improvements over baseline schemes based on single-user codes.

This paper investigates low-latency streaming codes for a three-node relay network. The source transmits a sequence of messages (streaming messages) to the destination through the relay between them, where the first-hop channel from the source to the relay and the second-hop channel from the relay to the destination are subject to packet erasures. Every source message must be recovered perfectly at the destination subject to a maximum decoding delay of T time slots, meaning that a message generated at a time slot must be recovered within the subsequent T time slots. In any sliding window of T +1 time slots, we assume no more than N1 and N2 erasures are introduced by the first-hop channel and second-hop channel respectively. Under this channel loss assumption, we fully characterize the maximum achievable rate in terms of T, N1 and N2.

Multi-user MIMO (MU-MIMO) is a technique in 802.11ac and 802.11ax that improves spectral efficiency by allowing concurrent communication between one AP and multiple clients. In practice, the expected gain is not always achieved and is sometimes even negative. Using a commodity 802.11ac AP, we experimentally determine that the inclusion of clients either in motion or with low SNR can cause throughput below that of single-user transmissions. We then propose a pre-screening algorithm using reinforcement learning to predict if a client can benefit from participating in MU-MIMO. Our algorithm is based on a sequence of channel state information (CSI), SNR, and client device type, and can automatically adapt to the motion of individual clients. Experimental results using a commodity AP show that the additional implementation of the pre-screening algorithm alone, without otherwise modifying MU-MIMO client grouping or link parameter selection algorithms, can improve system throughput by up to 40% when half of the clients are moving. Over 20% throughput improvement is maintained when between 25% to 75% of the clients are moving.

This paper considers multiplexing two sequences of messages with two different decoding delays over a packet erasure channel. In each time slot, the source constructs a packet based on the current and previous messages and transmits the packet, which may be erased when the packet travels from the source to the destination. The destination must perfectly recover every source message in the first sequence subject to a decoding delay ${T}_{\mathrm{ v}}$ and every source message in the second sequence subject to a shorter decoding delay ${T}_{\mathrm{ u}}\le {T}_{\mathrm{ v}}$ . We assume that the channel loss model introduces a burst erasure of a fixed length B on the discrete timeline. Under this channel loss assumption, the capacity region for the case where ${T}_{\mathrm{ v}}\le {T}_{\mathrm{ u}}+{B}$ was previously solved. In this paper, we fully characterize the capacity region for the remaining case ${T}_{\mathrm{ v}}> {T}_{\mathrm{ u}}+{B}$ . The key step in the achievability proof is achieving the non-trivial corner point of the capacity region through using a multiplexed streaming code constructed by superimposing two single-stream codes. The main idea in the converse proof is obtaining a genie-aided bound when the channel is subject to a periodic erasure pattern where each period consists of a length-B burst erasure followed by a length- ${T}_{\mathrm{ u}}$ noiseless duration.

We present a novel method that supports the learning of multiple classification tasks from geographically distributed data. By combining locally trained generative adversarial networks (GANs) with a small fraction of original data samples, our proposed scheme can train multiple discriminative models at a central location with low communication overhead. Experiments using common image datasets (MNIST, CIFAR10, LSUN-20, Celeb-A) show that our proposed scheme can achieve comparable classification accuracy as the ideal classifier trained using all data from all sites. We further demonstrate that our method can scale to 10 sites without sacrificing classification accuracy for large datasets such as LSUN-20.