Voice communications are valued for their ease of use and the rich information they provide, offering an immediate, clear, and efficient way to convey messages. However, ensuring the clarity and reliability of voice communications in low-bandwidth networks poses a technical challenge. This research explores the efficacy of Text-to-Speech (TTS) models and vocoder combinations versus traditional audio codecs in low-bandwidth networks, highlighting considerations for voice clarity and network resource management. Traditional audio codecs in bandwidth-limited environments often compromise audio quality and reliability. On the contrary, TTS models, supported by the advancements in deep and machine learning, present a potential alternative. Through a methodical comparison using various evaluation metrics, the study aims to offer valuable insights into their comparative impacts on audio quality and network behavior.

Augmented Reality (AR) enables the transmission of intent using the physical area in which humans and robots interact as a shared canvas. Studies exploring AR for human-robot collaboration have reported mixed findings on the relationship between cognitive workload and task performance. In this study, we developed an AR user interface (UI) that guides the user to perform a pick-and-place task while collaborating with a robot. A repeated measures mixed-methods study with sixteen participants demonstrated that AR UI significantly impacted task performance, where users traveled longer distances at slower speeds to pick-and-place objects than the control group. Additionally, UI significantly impacted the cognitive load, where participants reported higher NASA-TLX scores while using AR UI. Finally, users reported significantly lower situational awareness and low usability scores while using AR UI. Our findings suggest that the AR UI negatively impacts human-robot collaboration, calling for further investigation.

Monitoring and predicting user task performance is critical as it provides valuable insights for developing personalized human-system interactions. Key factors that impact task performance include cognitive workload, physiological responses, and affective states. However, a lack of consideration of any of these factors could lead to inaccurate task performance prediction because of their interplay. To address this challenge, we developed a novel hierarchical machine learning approach that considers these three factors to predict task performance. We exposed twenty-eight participants to a two-step experimental study. The first step aimed to induce different affective states using a validated video database. The second step required participants to perform validated low and high cognitive workload-inducing tasks. To evaluate the performance, we compared the models developed using the hierarchical approach that uses emotional and physiological information, to models that use only physiological information. We observed that our proposed approach always outperformed the models that only use physiological information to predict task performance by achieving a better average person independent mean absolute error. However, information gained across various models using the hierarchical approach was not linear. Additionally, we found that the top predictors for each model varied, and the model with the highest information gain included emotional features. These findings suggest the importance of choosing the appropriate machine learning model and predictors for building robust models for predicting task performance.

In recent years, robots have become vital to achieving manufacturing competitiveness. Especially in industrial environments, a strong level of interaction is reached when humans and robots form a dynamic system that works together towards achieving a common goal or accomplishing a task. However, the human-robot collaboration can be cognitively demanding, potentially contributing to cognitive fatigue. Therefore, the consideration of cognitive fatigue becomes particularly important to ensure the efficiency and safety in the overall human-robot collaboration. Additionally, sex is an inevitable human factor that needs further investigation for machine learning model development given the perceptual and physiological differences between the sexes in responding to fatigue. As such, this study explored sex differences and labeling strategies in the development of machine learning models for cognitive fatigue detection. Sixteen participants, balanced by sex, recruited to perform a surface finishing task with a UR10 collaborative robot under fatigued and non-fatigued states. Fatigue perception and heart rate activity data collected throughout to create a dataset for cognitive fatigue detection. Equitable machine learning models developed based on perception (survey responses) and condition (fatigue manipulation). The labeling approach had a significant impact on the accuracy and F1-score, where perception-based labels lead to lower accuracy and F1-score for females likely due to sex differences in reporting of fatigue. Additionally, we observed a relationship between heart rate, algorithm type, and labeling approach, where heart rate was the most significant predictor for the two labeling approaches and for all the algorithms utilized. Understanding the implications of label type, algorithm type, and sex on the design of fatigue detection algorithms is essential to designing equitable fatigue-adaptive human-robot collaborations across the sexes.

Convenient solutions that cater to different types of applications always gather more users. LoRa, one of the solutions in the IoT space is quite flexible at the physical layer, but link layer LoRa protocols lack similar flexibility. As a result, their performance remains suitable for the exchange of short messages in sensor networks, but does not extend to real-time and streaming IoT applications. Beartooth Relay Protocol (BRP) aims to address these limitations with a configurable link layer solution without compromising performance to sustain real-time flows over LoRa. We demonstrate that BRP expands the performance envelope of LoRa and makes it a suitable PHY technology for a wide range of IoT applications.

We present a framework for performance optimization in serverless edge-cloud platforms using dynamic task placement. We focus on applications for smart edge devices, for example, smart cameras or speakers, that need to perform processing tasks on input data in real to near-real time. Our framework allows the user to specify cost and latency requirements for each application task, and for each input, it determines whether to execute the task on the edge device or in the cloud. Further, for cloud executions, the framework identifies the container resource configuration needed to satisfy the performance goals. We have evaluated our framework in simulation using measurements collected from serverless applications in AWS Lambda and AWS Greengrass. In addition, we have implemented a prototype of our framework that runs in these same platforms. In experiments with our prototype, our models can predict average end-to-end latency with less than 6% error, and we obtain almost three orders of magnitude reduction in end-to-end latency compared to edge-only execution.

HTTP/2 (h2) was standardized in 2015 as an improvement to HTTP/1.1 (h1) to achieve faster webpage load times (PLTs) [5]. Previous studies have shown both improvement and degradation in PLT when using h2 with respect to h1 [6, 8]. The disagreement about h2 performance from these studies motivates further investigation as to whether and under what conditions h2 brings the performance benefits that were originally envisioned [5].

The transition to IPv6 cellular networks creates uncertainty for content providers (CPs) and content delivery networks (CDNs) of whether and how to follow suit. Do CPs that update their CDN contracts to allow IPv6 hosting achieve better, or worse performance in mobile networks? Should CDNs continue to host mobile content over IPv4 networks, or persuade to their CP customers the performance benefits of IPv6 content delivery?

Interactive real-time communication between people and machine enables innovations in transportation, health care, etc. Using voice or gesture commands improves usability and broad public appeal of such systems. In this paper we experimentally evaluate Google speech recognition and Apple Siri - two of the most popular cloud-based speech recognition systems. Our goal is to evaluate the performance of these systems under different network conditions in terms of command recognition accuracy and round trip delay - two metrics that affect interactive application usability. Our results show that speech recognition systems are affected by loss and jitter, commonly present in cellular and WiFi networks. Finally, we propose and evaluate a network coding transport solution to improve the quality of voice transmission to cloud-based speech recognition systems. Experiments show that our approach improves the accuracy and delay of cloud speech recognizers under different loss and jitter values.

Modern websites use Content Delivery Networks (CDNs) to speed up the delivery of static content. However, we show that DNS-based selection of CDN servers can be refined to fully deliver on the speedup of CDNs. We propose DNS-Proxy (dp), a client-side process that shares load-balancing functionality with CDNs by choosing from among resolved CDN servers based on last mile network performance. Our measurement study of CDN infrastructure deployed by five major CDN providers shows that dp reduces webpage load time by 29% on average. If dp has already resolved the domain, the reduction in webpage load time is as much as 40%. Finally, dp reduces the load time of individual static Web objects by as much as 43%. We argue that dp enables a more effective use of existing content delivery infrastructure and represents a complementary strategy to a continual increase of geographic content availability.

Interactive mobile Internet applications have become increasingly common and necessary to perform everyday tasks such as augmented reality, online gaming, video chat, and cloud-based voice recognition. These applications speed up their communications by connecting users to the closest servers and clustering nearby users together. In replica server selection and client clustering, determination of the closest host quickly and accurately is crucial to interactive application responses and the satisfaction of users' expectations. Researchers commonly use latency as the primary metric of network proximity and have developed various latency approximation tools. However, these tools do not yet offer an attractive balance of measurement accuracy, scalability, and maintainability. In this paper, we propose a new latency estimation system for arbitrary hosts using host-to-CDN latency measurements. Compared to existing latency estimation tools, our technique offers superior coverage of the IP address space and latency estimation accuracy. With improved coverage and accuracy of latency estimation it will become easier to establish low latency connections between hosts in a network, improving the responsiveness of interactive Internet applications.

Assessing multi-hop interpersonal trust in online social networks (OSNs) is critical for many social network applications such as online marketing but challenging due to the difficulties of handling complex OSN topology, in existing models such as subjective logic, and the lack of effective validation methods. To address these challenges, we for the first time properly define trust propagation and combination in arbitrary OSN topologies by proposing 3VSL (Three-Valued Subjective Logic). The 3VSL distinguishes the posteriori and priori uncertainties existing in trust, and the difference between distorting and original opinions, thus be able to compute multi-hop trusts in arbitrary graphs. We theoretically proved the capability based on the Dirichlet distribution. Furthermore, an online survey system is implemented to collect interpersonal trust data and validate the correctness and accuracy of 3VSL in real world. Both experimental and numerical results show that 3VSL is accurate in computing interpersonal trust in OSNs.

The use of swarm robotics in search tasks is an active area of research. A variety of algorithms have been developed that effectively direct robots toward a desired target by leveraging their collaborative sensing capabilities. Unfortunately, these algorithms often neglect the task of communicating possible task solutions outside of the swarm. Many scenarios require a monitoring station that must receive updates from robots within the swarm. This task is trivial in constrained locations, but becomes difficult as the search area increases and communication between nodes is not always possible. A second shortcoming of existing algorithms is the inability to find and track mobile targets. We propose an extension to the distributed Particle Swarm Optimization algorithm that is both communication-aware and capable of tracking mobile targets within a search space. Simulated experiments show that our algorithm returns more accurate solutions to a monitoring station than existing algorithms, especially in scenarios, where the target value or location changes over time.

Online Social Networks (OSN) are fun, popular, and socially significant. An integral part of their success is the immense size of their global user base. To provide a consistent service to all users, Facebook, the world’s largest OSN, is heavily dependent on centralized U.S. data centers, which renders service outside of the U.S. sluggish and wasteful of Internet bandwidth. In this paper, we investigate the detailed causes of these two problems and identify mitigation opportunities. Because details of Facebook’s service remain proprietary, we treat the OSN as a black box and reverse engineer its operation from publicly available traces. We find that contrary to current wisdom, OSN state is amenable to partitioning and that its fine grained distribution and processing can significantly improve performance without loss in service consistency. Through simulations of reconstructed Facebook traffic over measured Internet paths, we show that user requests can be processed 79% faster and use 91% less bandwidth. We conclude that the partitioning of OSN state is an attractive scaling strategy for Facebook and other OSN services.

The availability of Internet services brings many benefits to developing regions, yet Internet deployment levels in these regions remain staggeringly low. In this work we investigate how existing cellular deployments, which have enjoyed more rapid and wider deployment than client Internet infrastructure, could be used to provide very low cost Internet services in underdeveloped rural areas. We propose a new service model in which traffic is delivered over multihop client-to-client connections that are coordinated by end-to-end control traffic exchanged over cellular infrastructure. To enable this scheme in low client density rural settings, we propose a novel data forwarding mechanism for opportunistic space-time paths. To explore multiple opportunistic paths, but without the high forwarding cost of replicating data on these paths, we use network coding and send only a fraction of the data on each path. Through extensive OPNET simulations we show that globally coordinated opportunistic forwarding enables service acceptable to most applications at only a fraction of cellular infrastructure load. We argue that the reduced load on the cellular infrastructure allows additional users to share services and cost of the network and has the potential to lower the per user price of data services in developing regions.

The rapid growth in the popularity of cellular networks has led to aggressive deployment and a rapid expansion of mobile services. Services based on the integration of cellular networks into the Internet have only recently become available, but are expected to become very popular. One current limitation to the deployment of many of these services is poor or unknown network performance, particularly in the cellular portion of the network. Our goal in this paper is to motivate and present the Mobile Internet Services Test (MIST) platform, a new distributed architecture to measure and characterize cellular network performance as experienced by mobile devices. We have used MIST to conduct preliminary measurements; evaluate MIST’s effectiveness; and motivate further measurement research.