Federated learning (FL) is a collaborative machine learning technique to train a global model (GM) without obtaining clients' private data. The main challenges in FL are statistical diversity among clients, limited computing capability among clients' equipment, and the excessive communication overhead between the server and clients. To address these challenges, we propose a novel sparse personalized FL scheme via maximizing correlation (FedMac). By incorporating an approximated $\ell _{1}$ -norm and the correlation between client models and GM into standard FL loss function, the performance on statistical diversity data is improved and the communicational and computational loads required in the network are reduced compared with nonsparse FL. Convergence analysis shows that the sparse constraints in FedMac do not affect the convergence rate of the GM, and theoretical results show that FedMac can achieve good sparse personalization, which is better than the personalized methods based on the $\ell _{2}$ -norm. Experimentally, we demonstrate the benefits of this sparse personalization architecture compared with the state-of-the-art personalization methods (e.g., FedMac, respectively, achieves 98.95%, 99.37%, 90.90%, 89.06%, and 73.52% accuracy on the MNIST, FMNIST, CIFAR-100, Synthetic, and CINIC-10 datasets under non-independent and identically distributed (i.i.d.) variants).

The mobile communication enabled by cellular networks is the one of the main foundations of our modern society. Optimizing the performance of cellular networks and providing massive connectivity with improved coverage and user experience has a considerable social and economic impact on our daily life. This performance relies heavily on the configuration of the network parameters. However, with the massive increase in both the size and complexity of cellular networks, network management, especially parameter configuration, is becoming complicated. The current practice, which relies largely on experts' prior knowledge, is not adequate and will require lots of domain experts and high maintenance costs. In this work, we propose a learning-based framework for handover parameter configuration. The key challenge, in this case, is to tackle the complicated dependencies between neighboring cells and jointly optimize the whole network. Our framework addresses this challenge in two ways. First, we introduce a novel approach to imitate how the network responds to different network states and parameter values, called auto-grouping graph convolutional network (AG-GCN). During the parameter configuration stage, instead of solving the global optimization problem, we design a local multi-objective optimization strategy where each cell considers several local performance metrics to balance its own performance and its neighbors. We evaluate our proposed algorithm via a simulator constructed using real network data. We demonstrate that the handover parameters our model can find, achieve better average network throughput compared to those recommended by experts as well as alternative baselines, which can bring better network quality and stability. It has the potential to massively reduce costs arising from human expert intervention and maintenance.

Model-based offline reinforcement learning (RL) aims to find highly rewarding policy, by leveraging a previously collected static dataset and a dynamics model. While the dynamics model learned through reuse of the static dataset, its generalization ability hopefully promotes policy learning if properly utilized. To that end, several works propose to quantify the uncertainty of predicted dynamics, and explicitly apply it to penalize reward. However, as the dynamics and the reward are intrinsically different factors in context of MDP, characterizing the impact of dynamics uncertainty through reward penalty may incur unexpected tradeoff between model utilization and risk avoidance. In this work, we instead maintain a belief distribution over dynamics, and evaluate/optimize policy through biased sampling from the belief. The sampling procedure, biased towards pessimism, is derived based on an alternating Markov game formulation of offline RL. We formally show that the biased sampling naturally induces an updated dynamics belief with policy-dependent reweighting factor, termed Pessimism-Modulated Dynamics Belief. To improve policy, we devise an iterative regularized policy optimization algorithm for the game, with guarantee of monotonous improvement under certain condition. To make practical, we further devise an offline RL algorithm to approximately find the solution. Empirical results show that the proposed approach achieves state-of-the-art performance on a wide range of benchmark tasks.

Invertible neural networks based on Coupling Flows (CFlows) have various applications such as image synthesis and data compression. The approximation universality for CFlows is of paramount importance to ensure the model expressiveness. In this paper, we prove that CFlows}can approximate any diffeomorphism in $C^k$-norm if its layers can approximate certain single-coordinate transforms. Specifically, we derive that a composition of affine coupling layers and invertible linear transforms achieves this universality. Furthermore, in parametric cases where the diffeomorphism depends on some extra parameters, we prove the corresponding approximation theorems for parametric coupling flows named Para-CFlows. In practice, we apply Para-CFlows as a neural surrogate model in contextual Bayesian optimization tasks, to demonstrate its superiority over other neural surrogate models in terms of optimization performance and gradient approximations.

Traffic engineering (TE) is a fundamental task in networking. Conventionally, traffic can take any path connecting the source and destination. Emerging technologies such as segment routing, however, use logical paths that are composed of shortest paths going through a predetermined set of middlepoints in order to reduce the flow table overhead of TE implementation. Inspired by this, in this paper, we introduce the problem of node-constrained TE, where the traffic must go through a set of middlepoints, and study its theoretical fundamentals. We show that the general node-constrained TE that allows the traffic to take any path going through one or more middlepoints is NP-hard for directed graphs but strongly polynomial for undirected graphs, unveiling a profound dichotomy between the two cases. We also investigate a variant of node-constrained TE that uses only shortest paths between middlepoints, and prove that the problem can now be solved in weakly polynomial time for a fixed number of middlepoints, which explains why existing work focuses on this variant. Yet, if we constrain the end-to-end paths to be acyclic, the problem can become NP-hard. An important application of our work concerns flow centrality, for which we are able to derive complexity results. Furthermore, we investigate the middlepoint selection problem in general node-constrained TE. We introduce and study group flow centrality as a solution concept, and show that it is monotone but not submodular. Our work provides a thorough theoretical treatment of node-constrained TE and sheds light on the development of the emerging node-constrained TE in practice.

Network failures are common in wide area networks (WANs). Failure recovery in a software-defined WAN takes minutes or longer, as the controller needs to calculate a new traffic engineering solution and update the forwarding rules across all switches. This severely degrades application performance. Existing reactive and proactive approaches inevitably lead to transient congestion or bandwidth underutilization and impair the efficiency of running the expensive WANs. We present Sentinel, a novel failure recovery system for traffic engineering in software-defined WANs. Sentinel pre-computes and installs backup tunnels to accelerate failure recovery. When a link fails, switches locally redirect traffic to backup tunnels and recover immediately in the data plane, thus substantially reducing the transient congestion compared to reactive rescaling. On the other hand, Sentinel completely avoids the bandwidth headroom required by existing proactive approaches. Extensive experiments on Mininet and numerical simulations show that similar to state-of-the-art FFC, Sentinel reduces congestion by 45% compared with rescaling, and its algorithm runs much faster than FFC. Sentinel only introduces a small number of additional forwarding rules and can be readily implemented on today’s Openflow switches.

Data center networks demand high-performance, robust, and practical data plane load balancing protocols. Despite progress, existing work falls short of meeting these requirements. We design, analyze, and evaluate Luopan, a novel sampling based load balancing protocol that overcomes these challenges. Luopan operates at flowcell granularity similar to Presto. It periodically samples a few paths for each destination switch and directs flowcells to the least congested one. By being congestion-aware, Luopan improves flow completion time (FCT), and is more robust to topological asymmetries compared to Presto. The sampling approach simplifies the protocol and makes it much more scalable for implementation in large-scale networks compared to existing congestion-aware schemes. We provide analysis to show that Luopan's periodic sampling has the same asymptotic behavior as instantaneous sampling: taking 2 random samples provides exponential improvements over 1 sample. We conduct comprehensive packet-level simulations with production workloads. The results show that Luopan consistently outperforms state-of-the-art schemes in large-scale topologies. Compared to Presto, Luopan with 2 samples improves the 99.9%ile FCT of mice flows by up to 35 percent, and average FCT of medium and elephant flows by up to 30 percent. Luopan also performs significantly better than Local Sampling with large asymmetry.

The inference of the causal relationship between a pair of observed variables is a fundamental problem in science, and most existing approaches are based on one single causal model. In practice, however, observations are often collected from multiple sources with heterogeneous causal models due to certain uncontrollable factors, which renders causal analysis results obtained by a single model skeptical. In this paper, we generalize the Additive Noise Model (ANM) to a mixture model, which consists of a finite number of ANMs, and provide the condition of its causal identifiability. To conduct model estimation, we propose Gaussian Process Partially Observable Model (GPPOM), and incorporate independence enforcement into it to learn latent parameter associated with each observation. Causal inference and clustering according to the underlying generating mechanisms of the mixture model are addressed in this work. Experiments on synthetic and real data demonstrate the effectiveness of our proposed approach.

Cellular network configuration is critical for network performance. Current practice is mostly based on field experience and manual adjustment. The process is labor-intensive, error-prone, and far from optimal. To automate and optimize cellular network configuration, in this paper, we propose an online-learning-based joint-optimization approach that addresses a few specific challenges: limited data availability, convoluted sample data, highly complex optimization due to interactions among neighboring cells, and the need to adapt to network dynamics. In our approach, to learn an appropriate utility function for a cell, we develop a neural-network-based model that addresses the convoluted sample data issue and achieves good accuracy based on data aggregation. Based on the utility function learned, we formulate a global network configuration optimization problem. To solve this high-dimensional nonconcave maximization problem, we design a Gibbs-sampling-based algorithm that converges to an optimal solution when a technical parameter is small enough. Furthermore, we design an online scheme that updates the learned utility function and solves the corresponding maximization problem efficiently to adapt to network dynamics. To illustrate the idea, we use the case study of pilot power configuration. Numerical results illustrate the effectiveness of the proposed approach.

IP traffic classification has been a vitally important topic that attracts persistent interest in the networking and machine learning communities for past decades. While there exist quite a number of works applying machine learning techniques to realize IP traffic classification, most works suffer from limitations like either heavily depending on handcrafted features or be only able to handle offline traffic classification. To get rid of the aforementioned weakness, in this paper, we propose our online Convolutional Neural Networks (CNNs) based traffic classification framework named Seq2Img. The basic idea is to employ a compact nonparametric kernel embedding based method to convert early flow sequences into images which fully capture the static and dynamic behaviors of different applications and avoid using handcrafted features that might cause loss of information. A CNN is then applied on the generated images to obtain traffic classification results. Experiments on real network traffic are conducted and encouraging results justify the efficacy of our proposed approach.

Cellular network configuration is critical for network performance. Current practice is labor-intensive, error-prone, and far from optimal. To automate efficient cellular network configuration, in this work, we propose an online-learning-based joint-optimization approach that addresses a few specific challenges: limited data availability, convoluted sample data, highly complex optimization due to interactions among neighboring cells, and the need to adapt to network dynamics. In our approach, to learn an appropriate utility function for a cell, we develop a neural-network-based model that addresses the convoluted sample data issue and achieves good accuracy based on data aggregation. Based on the utility function learned, we formulate a global network configuration optimization problem. To solve this high-dimensional non-concave maximization problem, we design a Gibbs-sampling-based algorithm that converges to an optimal solution when a technical parameter is small enough. Furthermore, we design an online scheme that updates the learned utility function and solves the corresponding maximization problem efficiently to adapt to network dynamics. To illustrate the idea, we use the case study of pilot power configuration. Numerical results illustrate the effectiveness of the proposed approach.

Kernel methods have been successfully applied to reinforcement learning problems to address some challenges such as high dimensional and continuous states, value function approximation and state transition probability modeling. In this paper, we develop an online policy search algorithm based on a recent state-of-the-art algorithm REPS-RKHS that uses conditional kernel embeddings. Our online algorithm inherits the advantages of REPS-RKHS, including the ability to learn non-parametric control policies for infinite horizon continuous MDPs with high- dimensional sensory representations. Different from the original REPS-RKHS algorithm which is based on batch learning, the proposed online algorithm updates the model in an online fashion and thus is able to capture and respond to rapid changes in the system dynamics. In addition, the online update operation takes constant time (i.e., independent of the sample size n), which is much more efficient computationally and allows the policy to be continuously revised. Experiments on different domains are conducted and results show that our online algorithm outperforms the original algorithm.

Network traffic volume estimation and prediction is an important research topic that attracts persistent attention from the networking community and the machine learning community. Although there has been extensive work on estimating or predicting the traffic matrix using time series models, low rank matrix decomposition et. al, to the best of our knowledge, there is few work investigating the problem whether we are able to estimate and predict the traffic volume based on some statistics of the traffic which are much less costly to collect, for example, the flow counts. In this paper, we propose to model the relationship between the traffic volume and simple statistics about flows using a Hidden Markov Model based on which we can avoid direct measurement of the traffic volume but instead we estimate and predict the hidden traffic volume based on those simple flow statistics which are collected by some sketch techniques. We demonstrate the feasibility and effectiveness of our proposed method using some semi-simulation and real data experimental results.

We describe an emerging application of data mining in the context of computer networks. This application concerns the problem of predicting the size of a flow and detecting elephant flows (very large flows). Flow size is a very important statistic that can be used to improve routing, load balancing and scheduling in computer networks. Flow size prediction is particularly challenging since flow patterns continuously change and predictions must be done in real time (milliseconds) to avoid delays. We describe how to formulate the problem as an online machine learning task to continuously adjust to changes in flow traffic. We evaluate the predictive nature of a set of features and the accuracy of three online predictors based on neural networks, Gaussian process regression and online Bayesian Moment Matching on three datasets of real traffic. We also demonstrate how to use such online predictors to improve routing (i.e., reduced flow completion time) in a network simulation.

Data center networks demand high-performance, robust, and practical data plane load balancing protocols. Despite progress, existing work falls short of satisfying these requirements. We design and evaluate Luopan, a novel sampling based load balancing protocol that overcomes these challenges. Luopan operates at flowcell granularity similar to Presto. It periodically samples a few paths to each destination switch and directs flowcells to the least congested one. By being congestion-aware, Luopan improves flow completion time (FCT), and is more robust to topological asymmetries compared to Presto. The sampling approach simplifies the protocol and makes it much more scalable for implementation in large-scale networks compared to existing congestion-aware schemes. We conduct comprehensive packet-level simulations with a production workload. The results show that Luopan consistently outperforms state-of-the-art schemes in large-scale symmetric and asymmetric topologies. Compared to Presto, Luopan with 2 samples improves the 99%ile FCT of mice flows by up to 45%, and average FCT of medium flows by ~20%.

We present Sentinel, a novel failure recovery system for traffic engineering that pre-computes and installs backup tunnels to improve the robustness of software defined wide area networks (WANs). When a link fails, switches locally redirect traffic to backup tunnels and recover immediately in the data plane, thus substantially reducing the transient congestion compared to reactive rescaling. On the other hand Sentinel completely avoids the bandwidth headroom required by existing proactive approaches like FFC, and improves efficiency of operating the expensive WAN. We make several technical contributions in designing Sentinel. We formulate traffic engineering with backup tunnels (TE-BT) as optimization programs. We propose an approximation algorithm to efficiently solve the problem. We further present a concrete design and implementation of the system based on Openflow group tables for backup tunnels. Extensive experiments on Mininet and numerical simulations show that similar to FFC, Sentinel reduces congestion by 45% compared with rescaling, and its algorithm runs much faster than FFC. Sentinel only introduces a small number of additional forwarding rules and can be readily implemented on today's Openflow switches.

Leveraging application-level requirements using coflows has recently been shown to improve application-level communication performance in data-parallel clusters. However, existing coflow-based solutions rely on modifying applications to extract coflows, making them inapplicable to many practical scenarios.

In this paper, we present our effort towards comprehensive traffic forecasting for big data applications using external, light-weighted file system monitoring. Our idea is motivated by the key observations that rich traffic demand information already exists in the log and meta-data files of many big data applications, and that such information can be readily extracted through run-time file system monitoring. As the first step, we use Hadoop as a concrete example to explore our methodology and develop a system called HadoopWatch to predict traffic demands of Hadoop applications. We further implement HadoopWatch in a small-scale testbed with 10 physical servers and 30 virtual machines. Our experiments over a series of MapReduce applications demonstrate that HadoopWatch can forecast the traffic demand with almost 100% accuracy and time advance. Furthermore, it makes no modification on the Hadoop framework, and introduces little overhead to the application performance. Finally, to showcase the utility of accurate traffic prediction made by HadoopWatch, we design and implement a simple HadoopWatch-enabled network optimization module into the HadoopWatch controller, and with realistic Hadoop job benchmarks we find that even a simple algorithm can leverage the forecasting results provided by HadoopWatch to significantly improve the Hadoop job completion time by up to 14.72%.

Data-parallel computing frameworks (DCF) such as MapReduce, Spark, and Dryad etc. Have tremendous applications in big data and cloud computing, and throw tons of flows into data center networks. In this paper, we design and implement FLOW PROPHET, a general framework to predict traffic flows for DCFs. To this end, we analyze and summarize the common features of popular DCFs, and gain a key insight: since application logic in DCFs is naturally expressed by directed acyclic graphs (DAG), DAG contains necessary time and data dependencies for accurate flow prediction. Based on the insight, FLOW PROPHET extracts DAGs from user applications, and uses the time and data dependencies to calculate flow information 4-tuple, (source, destination, flow size, establish time), ahead-of-time for all flows. We also provide generic programming interface to FLOW PROPHET, so that current and future DCFs can deploy FLOW PROPHET readily. We implement FLOW PROPHET on both Spark and Hadoop, and perform extensive evaluations on a testbed with 37 physical servers. Our implementation and experiments demonstrate that, with time in advance and minimal cost, FLOW PROPHET can achieve almost 100% accuracy in source, destination, and flow size predictions. With accurate prediction from FLOW PROPHET, the job completion time of a Hadoop TeraSort benchmark is reduced by 12.52% on our cluster with a simple network scheduler.

In the data flow models of today's data center applications such as MapReduce, Spark and Dryad, multiple flows can comprise a coflow group semantically. Only completing all flows in a coflow is meaningful to an application. To optimize application performance, routing and scheduling must be jointly considered at the level of a coflow rather than individual flows. However, prior solutions have significant limitation: they only consider scheduling, which is insufficient. To this end, we present Rapier, a coflow-aware network optimization framework that seamlessly integrates routing and scheduling for better application performance. Using a small-scale testbed implementation and large-scale simulations, we demonstrate that Rapier significantly reduces the average coflow completion time (CCT) by up to 79.30% compared to the state-of-the-art scheduling-only solution, and it is readily implementable with existing commodity switches.

In this paper, we study the performance of opportunistic scheduling in wireless networks from the perspective of information and entropy. In opportunistic scheduling, we allocate a limited number of channels to a certain number of nodes so as to maximize the network performance. Due to the inherent uncertainty of the system input represented by random variables with certain probability distributions, even under the optimal scheduling strategy, we may not achieve the best network performance. In our proposed model, we mathematically formulate the relationship between system uncertainty characterized by entropy and network performance, i.e., we give the lower and upper bounds of network performance with given entropy of the uncertain input. Based on this result, we can determine quantitatively the impact of system uncertainty on the performance of of opportunistic scheduling in wireless networks.

Information theory has made great impact on research and development of communication systems. However, research in networking, a system built on communication components, has not benefited much from information theory and a theoretical framework to guide the analysis and design of networks is still lacking. Therefore, in this paper, we propose an information-theoretical framework to explore the quantitative relationship between information and network performance. Another attribute of information, namely information quality (IQ), is proposed to reflect the degree of importance of information to the target performance metric. IQ can be utilized to identify important information, so that network resource can first be allocated to such information, and hence to improve the network performance. Moreover, information efficiency (IE) is proposed as a metric to measure the information efficiency of network protocols and it can be used to analyze and compare the performance of different network protocols. Opportunistic scheduling is used to illustrate our proposed framework.

Many wireless network protocols have to deal with inaccurate information due to the lack of sufficient knowledge of the scenario or other limitations. Therefore, in this paper, we study information accuracy and investigate its quantitative impact on wireless network performance. First, we introduce an entropy-performance framework to model the relationship between wireless network performance and entropy, which characterizes the uncertainty of the input. Under this framework, we quantify the performance variations due to the availability of side information and find that the system performance improvement has a positive linear relationship with the amount of mutual information between input and side information. Subsequently, information accuracy is proposed to reflect the reliability of information. We show that information accuracy has a natural relationship with mutual information and we quantify the impact of information accuracy on wireless network performance.

In this paper, we study the performance of resource-constrained systems from the perspective of information theory. Such a system consists of many components which may contribute to the performance, but resource can only be allocated to some of them. We desire to allocate the limited resources effectively so as to maximize the system performance. Usually, we have incomplete information about the system or the system has inherent randomness. Even with the optimal allocation strategy according to the available (uncertain) information, we may not achieve the best system performance. We propose a model for the generic resource-constrained system and mathematically formulate the relationship between system uncertainty, characterized by entropy, and performance. Based on this result, we can determine how the system uncertainty quantitatively influences the performance. Examples of applications of the model to data storage and peer-to-peer file sharing are also given.

This paper develops a novel mechanism for traffic engineering, which combines differentiated services (DiffServ) and constraint-based routing together to meet the different QoS requirements, while preserving load balancing simultaneously. Simulation results illustrate the advantage of this mechanism. The proposed MPLS-TE mechanism can guarantee QoS requirements of the traffic flows and outperforms the traditional methods in terms of load balancing and throughout, so that the capability of the network is improved in evidence.