James Gross

Vishnu Narayanan Moothedath, Jaya Prakash Champati, James Gross

We consider a resource-constrained Edge Device (ED), such as an IoT sensor or a microcontroller unit, embedded with a small-size ML model (S-ML) for a generic classification application and an Edge Server (ES) that hosts a large-size ML model (L-ML). Since the inference accuracy of S-ML is lower than that of the L-ML, offloading all the data samples to the ES results in high inference accuracy, but it defeats the purpose of embedding S-ML on the ED and deprives the benefits of reduced latency, bandwidth savings, and energy efficiency of doing local inference. In order to get the best out of both worlds, i.e., the benefits of doing inference on the ED and the benefits of doing inference on ES, we explore the idea of Hierarchical Inference (HI), wherein S-ML inference is only accepted when it is correct, otherwise the data sample is offloaded for L-ML inference. However, the ideal implementation of HI is infeasible as the correctness of the S-ML inference is not known to the ED. We propose an online meta-learning framework that the ED can use to predict the correctness of the S-ML inference. In particular, we propose to use the maximum softmax value output by S-ML for a data sample and decide whether to offload it or not. The resulting online learning problem turns out to be a Prediction with Expert Advice (PEA) problem with continuous expert space. We propose two different algorithms and prove sublinear regret bounds for them without any assumption on the smoothness of the loss function. We evaluate and benchmark the performance of the proposed algorithms for image classification application using four datasets, namely, Imagenette and Imagewoof, MNIST, and CIFAR-10.

Mattia Sartori, Chetna Singhal, Neelabhro Roy et al.

The miniaturisation of sensors and processors, the advancements in connected edge intelligence, and the exponential interest in Artificial Intelligence are boosting the affirmation of autonomous nano-size drones in the Internet of Robotic Things ecosystem. However, achieving safe autonomous navigation and high-level tasks such as exploration and surveillance with these tiny platforms is extremely challenging due to their limited resources. This work focuses on enabling the safe and autonomous flight of a pocket-size, 30-gram platform called Crazyflie 2.1 in a partially known environment. We propose a novel AI-aided, vision-based reactive planning method for obstacle avoidance under the ambit of Integrated Sensing, Computing and Communication paradigm. We deal with the constraints of the nano-drone by splitting the navigation task into two parts: a deep learning-based object detector runs on the edge (external hardware) while the planning algorithm is executed onboard. The results show the ability to command the drone at $\sim8$ frames-per-second and a model performance reaching a COCO mean-average-precision of $60.8$. Field experiments demonstrate the feasibility of the solution with the drone flying at a top speed of $1$ m/s while steering away from an obstacle placed in an unknown position and reaching the target destination. The outcome highlights the compatibility of the communication delay and the model performance with the requirements of the real-time navigation task. We provide a feasible alternative to a fully onboard implementation that can be extended to autonomous exploration with nano-drones.

Vishnu Narayanan Moothedath, Umang Agarwal, Umeshraja N et al.

We focus on a binary classification problem in an edge intelligence system where false negatives are more costly than false positives. The system has a compact, locally deployed model, which is supplemented by a larger, remote model, which is accessible via the network by incurring an offloading cost. For each sample, our system first uses the locally deployed model for inference. Based on the output of the local model, the sample may be offloaded to the remote model. This work aims to understand the fundamental trade-off between classification accuracy and the offloading costs within such a hierarchical inference (HI) system. To optimise this system, we propose an online learning framework that continuously adapts a pair of thresholds on the local model's confidence scores. These thresholds determine the prediction of the local model and whether a sample is classified locally or offloaded to the remote model. We present a closed-form solution for the setting where the local model is calibrated. For the more general case of uncalibrated models, we introduce H2T2, an online two-threshold hierarchical inference policy, and prove it achieves sublinear regret. H2T2 is model-agnostic, requires no training, and learns during the inference phase using limited feedback. Simulations on real-world datasets show that H2T2 consistently outperforms naive and single-threshold HI policies, sometimes even surpassing offline optima. The policy also demonstrates robustness to distribution shifts and adapts effectively to mismatched classifiers.

M. Olguín Muñoz, R. Klatzky, J. Wang et al.

Wearable Cognitive Assistants (WCA) are anticipated to become a widely-used application class, in conjunction with emerging network infrastructures like 5G that incorporate edge computing capabilities. While prototypical studies of such applications exist today, the relationship between infrastructure service provisioning and its implication for WCA usability is largely unexplored despite the relevance that these applications have for future networks. This paper presents an experimental study assessing how WCA users react to varying end-to-end delays induced by the application pipeline or infrastructure. Participants interacted directly with an instrumented task-guidance WCA as delays were introduced into the system in a controllable fashion. System and task state were tracked in real time, and biometric data from wearable sensors on the participants were recorded. Our results show that periods of extended system delay cause users to correspondingly (and substantially) slow down in their guided task execution, an effect that persists for a time after the system returns to a more responsive state. Furthermore, the slow-down in task execution is correlated with a personality trait, neuroticism, associated with intolerance for time delays. We show that our results implicate impaired cognitive planning, as contrasted with resource depletion or emotional arousal, as the reason for slowed user task executions under system delay. The findings have several implications for the design and operation of WCA applications as well as computational and communication infrastructure, and additionally for the development of performance analysis tools for WCA.

Henrik Forssell, Ragnar Thobaben, Hussein Al-Zubaidy et al.

We study the detection and delay performance impacts of a feature-based physical layer authentication (PLA) protocol in mission-critical machine-type communication (MTC) networks. The PLA protocol uses generalized likelihood-ratio testing based on the line-of-sight (LOS), single-input multiple-output channel-state information in order to mitigate impersonation attempts from an adversary node. We study the detection performance, develop a queueing model that captures the delay impacts of erroneous decisions in the PLA (i.e., the false alarms and missed detections), and model three different adversary strategies: data injection, disassociation, and Sybil attacks. Our main contribution is the derivation of analytical delay performance bounds that allow us to quantify the delay introduced by PLA that potentially can degrade the performance in mission-critical MTC networks. For the delay analysis, we utilize tools from stochastic network calculus. Our results show that with a sufficient number of receive antennas (approx. 4-8) and sufficiently strong LOS components from legitimate devices, PLA is a viable option for securing mission-critical MTC systems, despite the low latency requirements associated to corresponding use cases. Furthermore, we find that PLA can be very effective in detecting the considered attacks, and in particular, it can significantly reduce the delay impacts of disassociation and Sybil attacks.

Sahar Imtiaz, Hadi Ghauch, M. Mahboob Ur Rahman et al.

Explosive growth in the use of smart wireless devices has necessitated the provision of higher data rates and always-on connectivity, which are the main motivators for designing the fifth generation (5G) systems. To achieve higher system efficiency, massive antenna deployment with tight coordination is one potential strategy for designing 5G systems, but has two types of associated system overhead. First is the synchronization overhead, which can be reduced by implementing a cloud radio access network (CRAN)-based architecture design, that separates the baseband processing and radio access functionality to achieve better system synchronization. Second is the overhead for acquiring channel state information (CSI) of the users present in the system, which, however, increases tremendously when instantaneous CSI is used to serve high-mobility users. To serve a large number of users, a CRAN system with a dense deployment of remote radio heads (RRHs) is considered, such that each user has a line-of-sight (LOS) link with the corresponding RRH. Since, the trajectory of movement for high-mobility users is predictable; therefore, fairly accurate position estimates for those users can be obtained, and can be used for resource allocation to serve the considered users. The resource allocation is dependent upon various correlated system parameters, and these correlations can be learned using well-known \emph{machine learning} algorithms. This paper proposes a novel \emph{learning-based resource allocation scheme} for time division duplex (TDD) based 5G CRAN systems with dense RRH deployment, by using only the users' position estimates for resource allocation, thus avoiding the need for CSI acquisition. This reduces the overall system overhead significantly, while still achieving near-optimal system performance; thus, better (effective) system efficiency is achieved. (See the paper for full abstract)