Domain-Specific On-Device Object Detection Method

Object detection is a significant activity in computer vision, and various approaches have been proposed to detect varied objects using deep neural networks (DNNs). However, because DNNs are computation-intensive, it is difficult to apply them to resource-constrained devices. Here, we propose an on-device object detection method using domain-specific models. In the proposed method, we define object of interest (OOI) groups that contain objects with a high frequency of appearance in specific domains. Compared with the existing DNN model, the layers of the domain-specific models are shallower and narrower, reducing the number of trainable parameters; thus, speeding up the object detection. To ensure a lightweight network design, we combine various network structures to obtain the best-performing lightweight detection model. The experimental results reveal that the size of the proposed lightweight model is 21.7 MB, which is 91.35% and 36.98% smaller than those of YOLOv3-SPP and Tiny-YOLO, respectively. The f-measure achieved on the MS COCO 2017 dataset were 18.3%, 11.9% and 20.3% higher than those of YOLOv3-SPP, Tiny-YOLO and YOLO-Nano, respectively. The results demonstrated that the lightweight model achieved higher efficiency and better performance on non-GPU devices, such as mobile devices and embedded boards, than conventional models.

Privacy-Enhanced MQTT Protocol for Massive IoT

The growing expectations for ubiquitous sensing have led to the integration of countless embedded sensors, actuators, and RFIDs in our surroundings. Combined with rapid developments in high-speed wireless networks, these resource-constrained devices are paving the road for the Internet-of-Things paradigm, a computing model aiming to bring together millions of heterogeneous and pervasive elements. However, it is commonly accepted that the Privacy consideration remains one of its main challenges, a notion that does not only encompasses malicious individuals but can also be extended to honest-but-curious third-parties. In this paper, we study the design of a privacy-enhanced communication protocol for lightweight IoT devices. Applying the proposed approach to MQTT, a highly popular lightweight publish/subscribe communication protocol prevents no valuable information from being extracted from the messages flowing through the broker. In addition, it also prevents partners re-identification. Starting from a privacy-ideal, but unpractical, exact transposition of the Oblivious Transfer (OT) technology to MQTT, this paper follows an iterative process where each previous model’s drawbacks are appropriately mitigated all the while trying to preserve acceptable privacy levels. Our work provides resistance to statistical analysis attacks and dynamically supports new client participation. Additionally the whole proposal is based on the existence of a non-communicating 3rd party during pre-development. This particular contribution reaches a proof-of-concept stage through implementation, and achieves its goals thanks to OT’s indistinguishability property as well as hash-based topic obfuscations.

Load Classification: A Case Study for Applying Neural Networks in Hyper-Constrained Embedded Devices

The application of Artificial Intelligence to the industrial world and its appliances has recently grown in popularity. Indeed, AI techniques are now becoming the de-facto technology for the resolution of complex tasks concerning computer vision, natural language processing and many other areas. In the last years, most of the the research community efforts have focused on increasing the performance of most common AI techniques—e.g., Neural Networks, etc.—at the expenses of their complexity. Indeed, many works in the AI field identify and propose hyper-efficient techniques, targeting high-end devices. However, the application of such AI techniques to devices and appliances which are characterised by limited computational capabilities, remains an open research issue. In the industrial world, this problem heavily targets low-end appliances, which are developed focusing on saving costs and relying on—computationally—constrained components. While some efforts have been made in this area through the proposal of AI-simplification and AI-compression techniques, it is still relevant to study which available AI techniques can be used in modern constrained devices. Therefore, in this paper we propose a load classification task as a case study to analyse which state-of-the-art NN solutions can be embedded successfully into constrained industrial devices. The presented case study is tested on a simple microcontroller, characterised by very poor computational performances—i.e., FLOPS –, to mirror faithfully the design process of low-end appliances. A handful of NN models are tested, showing positive outcomes and possible limitations, and highlighting the complexity of AI embedding.

IMPROVED ROUTING PROTOCOLS BASED ON RPL FOR FULL IOT-WSN STACK

The Internet Engineering Task Force (IETF) standardized several protocols such as Constraint Application Protocol (CoAP) to run over WSN-IoT constrained devices. IPv6 is used to transmit packets over IEEE802.15.4 radio link called 6LoWPAN. The routing protocol for low power and lossy (RPL) network enable connectivity of WSN over IoT. Nodes in RPL optimized its path using objective function (OF), which depends on different node/link metrics. In this paper, the performance of IoT-WSN stack consisting of CoAP, 6LoWPAN and two proposed protocols based on RPL are evaluated. The two proposed protocols depend on a new OF based on link reliability and energy metric. The first is a modification of recently introduced protocol called Energy Threshold RPL (ETRPL) protocol. ETRPL uses the remaining energy of the preferred parent node as a metric. The other protocol used a new metric that combines energy consumption with ETX for all paths to the root called EERPL. The protocols are tested with full IoT-WSN stack and implemented using Cooja simulator. The results showed that ETRPL and EERPL performed better than standard RPL in terms of the energy consumption, average time delay, packet reception ratio, throughput and the number of dead nodes.

Ecc-bAsed ScalablE Revocation(EASER) scheme for selective sharing in resource-constrained devices

The full text of this preprint has been withdrawn by the authors due to author disagreement with the posting of the preprint. Therefore, the authors do not wish this work to be cited as a reference. Questions should be directed to the corresponding author.