The advent of tiny artificial intelligence (AI) accelerators enables AI to run at the extreme edge, offering reduced latency, lower power cost, and improved privacy. When integrated into wearable devices, these accelerators open exciting opportunities, allowing various AI apps to run directly on the body. We present Synergy that provides AI apps with besteffort performance via system-driven holistic collaboration over AI accelerator-equipped wearables. To achieve this, Synergy provides device-agnostic programming interfaces to AI apps, giving the system visibility and controllability over the app's resource use. Then, Synergy maximizes the inference throughput of concurrent AI models by creating various execution plans for each app considering AI accelerator availability and intelligently selecting the best set of execution plans. Synergy further improves throughput by leveraging parallelization opportunities over multiple computation units. Our evaluations with 7 baselines and 8 models demonstrate that, on average, Synergy achieves a 23.0× improvement in throughput, while reducing latency by 73.9% and power consumption by 15.8%, compared to the baselines

Overlapping cameras offer exciting opportunities to view a scene from different angles, allowing for more advanced, comprehensive and robust analysis. However, existing video analytics systems for multi-camera streams are mostly limited to (i) per-camera processing and aggregation and (ii) workload-agnostic centralized processing architectures. In this paper, we present Argus, a distributed video analytics system withcross-camera collaborationon smart cameras. We identify multi-camera, multi-target tracking as the primary task of multi-camera video analytics and develop a novel technique that avoids redundant, processing-heavy identification tasks by leveraging object-wise spatio-temporal association in the overlapping fields of view across multiple cameras. We further develop a set of techniques to perform these operations across distributed cameras without cloud support at low latency by (i) dynamically ordering the camera and object inspection sequence and (ii) flexibly distributing the workload across smart cameras, taking into account network transmission and heterogeneous computational capacities. Evaluation of three real-world overlapping camera datasets with two Nvidia Jetson devices shows that Argus reduces the number of object identifications and end-to-end latency by up to 7.13× and 2.19× (4.86× and 1.60× compared to the state-of-the-art), while achieving comparable tracking quality.

We present SensiX++, a multi-tenant runtime for adaptive model execution with integrated MLOps on edge devices, e.g., a camera, a microphone, or IoT sensors. SensiX++ operates on two fundamental principles: highly modular componentisation to externalise data operations with clear abstractions and document-centric manifestation for system-wide orchestration. First, a data coordinator manages the lifecycle of sensors and serves models with correct data through automated transformations. Next, a resource-aware model server executes multiple models in isolation through model abstraction, pipeline automation, and feature sharing. An adaptive scheduler then orchestrates the best-effort executions of multiple models across heterogeneous accelerators, balancing latency and throughput. Finally, microservices with REST APIs serve synthesised model predictions, system statistics, and continuous deployment. Collectively, these components enable SensiX++ to serve multiple models efficiently with fine-grained control on edge devices while minimising data operation redundancy, managing data and device heterogeneity, and reducing resource contention. We benchmark SensiX++ with 10 different vision and acoustics models across various multi-tenant configurations on different edge accelerators (Jetson AGX and Coral TPU) designed for sensory devices. We report on the overall throughput and quantified benefits of various automation components of SensiX++ and demonstrate its efficacy in significantly reducing operational complexity and lowering the effort to deploy, upgrade, reconfigure, and serve embedded models on edge devices.

No abstract available.

No abstract available.

We present eSense - an open and multi-sensory in-ear wearable platform for personal-scale behaviour analytics. eSense is a true wireless stereo (TWS) earbud and supports dual-mode Bluetooth and Bluetooth Low Energy. It is also augmented with a 6-axis in-ertial measurement unit and a microphone. We demonstrate the eSense platform, the data exploration tool with the open APIs for the real-time visualisation of multi-modal sensory data, and its manifestation in a 360° workplace well-being application.

Push notifications have become a prevalent way of communication for user devices to receive updates about their online profiles. State-of-the-art mechanisms for push notification delivery do not take users' spatio-temporal context into account as it would require a user device to retrieve the location and deliver it to the app servers, both of which are costly in terms of energy and traffic. In this paper, we provide an opportunistic, local, network-centric mechanism, WiPush, to deliver contextual notifications, leveraging the high density of WiFi access points (AP). WiPush does not require association with APs and uses WiFi public action frames to deliver notifications without abusing the underlying standards. Our evaluation shows that while WiPush can deliver short notifications to the users with high reliability without affecting the networking functionalities of the AP adversely, it can not be used to transfer large volumes of data.

This paper reports a study on the characterization of in-home Internet activity behavior based on application usage logs. We collected online activity data from 86 Belgium households for 60 days. We analyzed the activity traces to gain insights on the temporal traffic distribution, interaction regularity, and activity correlations. This analysis is then used to develop a generic method to segment households into designated groups showing similar behavioral profiles. Our technique combines interaction frequencies and regularities across activities for segmentation, and is able to reveal interesting time-slotted profile for each segment. These profiles aim to show the strength of routine behaviors in Internet usage, based on which we present a novel algorithm to predict future Internet activities of a household. Our algorithm shows that 60% of the households online activities can be predicted accurately 70% of times.

Industrial events and exhibitions play a powerful role in creating social relations amongst individuals and firms, enabling them to expand their social network so to acquire resources. However, often these events impose a spatial structure which impacts encounter opportunities. In this paper, we study the impact that the spatial configuration has on the formation of network relations. We designed, developed and deployed a Wi-Fi analytics solution comprising of wearable Wi-Fi badges and gateways in a large scale industrial exhibition event to study the spatio-temporal trajectories of the 2.5K+ attendees including two special groups: 34 investors and 27 entrepreneurs. Our results suggest that certain zones with designated functionalities play a key role in forming social ties across attendees and the different behavioural properties of investors and entrepreneurs can be explained through a spatial lens. Based on our findings we offer three concrete recommendations for future organisers of networking events.

WiFi-enabled buses and stops may form the backbone of a metropolitan delay tolerant network, that exploits nearby communications, temporary storage at stops, and predictable bus mobility to deliver non-real time information. This paper studies the problem of how to route data from its source to its destination in order to maximize the delivery probability by a given deadline. We assume to know the bus schedule, but we take into account that randomness, due to road traffic conditions or passengers boarding and alighting, affects bus mobility. We propose a simple stochastic model for bus arrivals at stops, supported by a study of real-life traces collected in a large urban network. A succinct graph representation of this model allows us to devise an optimal (under our model) single-copy routing algorithm and then extend it to cases where several copies of the same data are permitted. Through an extensive simulation study, we compare the optimal routing algorithm with three other approaches: minimizing the expected traversal time over our graph, minimizing the number of hops a packet can travel, and a recently-proposed heuristic based on bus frequencies. Our optimal algorithm outperforms all of them, but most of the times it essentially reduces to minimizing the expected traversal time. For values of deadlines close to the expected delivery time, the multi-copy extension requires only 10 copies to reach almost the performance of the costly flooding approach.

With the convergence between phone and data networks in LTE and 4G, cellular signaling traffic is increasingly carried over IP. Control plane functions, once performed by dedicated machinery, are evolving into large-scale network applications with strict requirements on delay, availability, and processing throughput as mandated by 3GPP standards. In this paper we present DMME, a distributed architecture that implements mobility management for next-generation cellular systems. DMME is a scalable and cost-effective drop-in replacement for the LTE mobility management entity (MME). We evaluate the DMME scheme via analysis and simulation under several deployment scenarios, using mobility patterns drawn from synthetic models and from traces collected in a production network. We also report preliminary performance figures by our DMME prototype implementation. Our results confirm that distributed architectures are a viable choice to reliably support high-throughput, latency-sensitive control plane functions.

Forwarding decisions in routing protocols rely on information about the destination nodes provided by routing table states. When paths to a destination change, corresponding states become invalid and need to be refreshed with control messages for resilient routing. In large and highly dynamic networks, this overhead can crowd out the capacity for data traffic. For such networks, we propose the concept of weak state, which is interpreted as a probabilistic hint, not as absolute truth. Weak state can remain valid without explicit messages by systematically reducing the confidence in its accuracy. Weak State Routing (WSR) is a novel routing protocol that uses weak state along with random directional walks for forwarding packets. When a packet reaches a node that contains a weak state about the destination with higher confidence than that held by the packet, the walk direction is biased. The packet reaches the destination via a sequence of directional walks, punctuated by biasing decisions. WSR also uses random directional walks for disseminating routing state and provides mechanisms for aggregating weak state. Our simulation results show that WSR offers a very high packet delivery ratio ( ≥ 98%). Control traffic overhead scales asO(N), and the state complexity is Θ(N3/2), whereNis the number of nodes. Packets follow longer paths compared to prior protocols (OLSR , GLS-GPSR , ), but the average path length is asymptotically efficient and scales asO(√N). Despite longer paths, WSR's end-to-end packet delivery delay is much smaller due to the dramatic reduction in protocol overhead.

State signaling and maintenance mechanisms play crucial roles in communication network protocols. State is used to facilitate indirections in protocols such as routing. Design approaches for traditional state signaling mechanisms have been categorized into soft and hard state. In both approaches, the state is deterministic. Hence, we call both as having strong state semantics, or more crisply, refer to them as strong state. If the state tracks entities with dynamic nature, strong state rapidly becomes invalidated and needs to be refreshed explicitly through control packets. In this paper, we evaluate the recently proposed weak state. Weak state is a generalization of soft state that is characterized by probabilistic semantics and local updates. It is interpreted as a probabilistic hint and not absolute truth. Weak state also contains the confidence in the state value, which is a measure of the probability that the state remains valid. The confidence or the state semantics is decayed locally without the need for explicit state update traffic traversing the network. The local updates also help the protocol use better estimates for the state value. We define two metrics, pure distortion and informed distortion, to evaluate the consistency of the weak state paradigm and compare it against strong state. Pure distortion measures the average gap between the actual value of the state and the value maintained at a remote node. On the other hand, the use of confidence increases the protocol's ability to cope with even large pure distortion. The resulting effective distortion is captured by the informed distortion metric. Using mathematical analysis, we compare weak with strong state. Local updates reduce the pure distortion because the protocol uses the best estimate of state value. The informed distortion is also significantly less because the probabilistic confidence value hints the protocol if the state is invalid. The weak state mechanism can be used to build protocols (eg: WSR [1]), which systematically interpret the state information. The state itself can be mostly updated locally, with less frequent explicit update messages over the network (i.e. leading to dramatic reductions in control traffic).

Routing in communication networks involves the indirection from a persistent name (or ID) to a locator and delivering packets based upon the locator. In a large-scale, highly dynamic network, the ID-to-locator mappings are both large in number, and change often. Traditional routing protocols require high overhead to keep these in directions up-to-date. In this paper, we propose Weak State Routing (WSR), a routing mechanism for large-scale highly dynamic networks. WSR's novelty is that it uses random directional walks biased occasionally by weak indirection state information in intermediate nodes. The indirection state information is weak, i.e. interpreted not as absolute truth, but as probabilistic hints. Nodes only have partial information about the region a destination node is likely to be. This method allows us to aggregate information about a number of remote locations in a geographic region. In other words, the state information maps a set-of-IDs to a it geographical region. The intermediate nodes receiving the random walk use a method similar to longest-prefix-match in order to prioritize their mappings to decide how to bias and forward the random walk. WSR can also be viewed as an unstructured distributed hashing technique. WSR displays good rare-object recall with scalability properties similar to structured DHTs, albeit with more tolerance to dynamism and without constraining the degree distribution of the underlying network.Through simulations, we show that WSR offers a high packet delivery ratio, more than 98%. The control packet overhead incurred in the network scales as O(N) for N-node networks. The number of mappings stored in the network appears to scale as Θ(N(3/2)). We compare WSR with Dynamic Source Routing (DSR) and geographic forwarding (GPSR) combined with Grid Location Service (GLS). Our results indicate that WSR delivers more packets with less overhead at the cost of increased path length.