Matteo Petracca

UTDrive: Driver Behavior and Speech Interactive Systems for In-Vehicle Environments

This paper describes an overview of the UTDrive project. UTDrive is part of an on-going international collaboration to collect and research rich multi-modal data recorded for modeling driver behavior for in-vehicle environments. The objective of the UTDrive project is to analyze behavior while the driver is interacting with speech-activated systems or performing common secondary tasks, as well as to better understand speech characteristics of the driver undergoing additional cognitive load. The corpus consists of audio, video, gas/brake pedal pressure, forward distance, GPS information, and CAN-Bus information. The resulting corpus, analysis, and modeling will contribute to more effective speech interactive systems with are less distractive and adjustable to the drivers cognitive capacity and driving situations.

T-Res: Enabling Reconfigurable In-network Processing in IoT-based WSNs

Internet of Things (IoT) standards are a key technology for enabling multi-purpose Wireless Sensor Networks (WSNs). Indeed, by providing a common interface for interacting with sensor nodes, they help decouple the sensing infrastructure from the applications running on top of it. To this end, much work has been done to define the semantics for controlling sensors and actuators. On the contrary, a solution for manipulating the in-network processing performed by an IoT-based WSN is still undefined. As a result, IoT applications usually run “out-of-the-network”, e.g., in high-end application servers or in the Cloud. This paper addresses such a limitation by proposing TRes, a new programming abstraction for IoT-based WSNs. The proposed solution allows using IoT protocols for changing the data-processing performed by sensor nodes and the interaction among them, thus enabling “in-network” applications. To prove the feasibility of T-Res, a real implementation based on Contiki OS is presented and evaluated. Evaluation results show that the use of T-Res, compared with the traditional “out-of-network” approach, lowers the risk of network congestion, reduces the power consumption, and can improve application response time.

Perceptual based voice multi-hop transmission over wireless sensor networks

Multimedia applications over wireless sensor networks (WSNs) are rapidly gaining interest by the research community in order to develop new and mission critical services such as environmental video monitoring and emergency speech calls. In this work we analyze the possibility of sending voice using a network of wireless tiny motes with the final goal of enhancing speech quality by protecting the most perceptually important packets. We first evaluate the speech quality for a modified version of the ITU-T G.711 standard implemented to fit the particular selected hardware. Hence, we propose a lowcomplexity measure to evaluate the perceptual importance of speech packets. When performing single-hop experimental data collection, we apply packet redundancy (protection) by using a cooperative mote (relay) which retransmits speech packets that are perceptually important to protect them against potential transmission losses. Collected experimental results are then used to assess multi-hop performance, showing that the combination of the selected hardware and the proposed perceptual marking algorithm achieves good speech quality levels, according to the MOS scale, while reducing the percentage of protected packets by 40% when compared to random protection.

Voice transmission over 802.11 wireless networks using analysis-by-synthesis packet classification

A new form of telephony is being made possible by the nearly-ubiquitous presence of 802.11 wireless local networks (WLANs). This paper presents a technique to improve speech quality for WLAN telephony. Speech compressed by the 3GPP GSM AMR coding standard at 12.2 kb/s is packetized and classified according to an analysis-by-synthesis estimate of the perceptual importance of each individual packet. Perceptually important packets are protected against noise and channel collisions by a simple form of forward error correction, i.e., packet repetition. Network simulations and perceptual quality measures have been used to evaluate the performance of the proposed technique. Preliminary results show that protecting the perceptually most important 10% of all speech packets provides the same performance delivered by randomly protecting twice as many packets.

BlueVoice: Voice communications over Bluetooth Low Energy in the Internet of Things scenario

Abstract Bluetooth Low Energy (Bluetooth LE) is a key technology in the envisioned Internet of Things (IoT) scenario. In fact, its extremely low-power characteristics make it one of the most suitable solutions to enable wireless communications among battery powered IoT objects ubiquitously deployed in the field with the aim of building smart environments. Although Bluetooth LE specification targets a specific set of applications mainly devoted to monitoring purposes, innovative solutions can lead to the adoption of such technology in different applications, such as multimedia streaming, allowing IoT objects to exploit new functionalities. In this direction this article presents BlueVoice, an application targeted to Bluetooth LE devices to enable speech streaming services. In the article BlueVoice is presented by first detailing the services set extension needed to support the new envisioned multimedia service, then a description of application choices is given, followed by an evaluation of its performance in real IoT objects. Thanks to the selected speech encoding technique, connection design choices and packetization strategies, BlueVoice application requires a communication bandwidth of 64.3 kbps to transmit audio at 16 kHz in ADPCM format. BlueVoice performance has been evaluated in terms of power consumption, memory and processing requirements, showing feasibility of the developed solution in resource constrained devices, thus confirming the correct choices in the application design. The set of performance information obtained show that BlueVoice is a viable solution to enable speech communications in ubiquitous wireless IoT nodes based on the Bluetooth LE technology.

Middleware solutions in WSN: The IoT oriented approach in the ICSI project

A large number of middleware solutions targeted to wireless sensor networks have been deployed during the years with the aim of bridging the gap between the high level requirements of the applications and the low level hardware complexity. In the paper, a review of state-of-the-art middleware platforms is first presented by highlighting basic design principles and network protocol solutions for next generation middleware targeted to the Internet of Things (IoT) scenario. Then, the general architecture of the middleware to be instantiated within the European project Intelligent Cooperative Sensing for Improved traffic efficiency (ICSI) is presented by detailing its architecture and IoT enabling solutions. The ICSI WSN middleware introduces state-of-the-art advances in IoT middleware design by merging a publish/subscribe model with a virtual machine based design.

Performance evaluation of FEC techniques based on BCH codes in video streaming over wireless sensor networks

Video streaming in wireless sensor networks is a promising and challenging application. The reduced amount of available bandwidth, as well as the low-computational power available for acquiring and processing video frames on tiny devices, imposes the transmission of low resolution video frames at a low frame rate. In such a scenario the amount of information carried by each video frame must be preserved as mush as possible in order to avoid possible artifacts in the reconstructed dynamics of the scene. In this paper we first evaluate the impact of the bit error rate on the quality of the received video stream, then we propose a forward error correction technique based on the use of BCH codes to preserve the video quality. The proposed technique, fully compliant with the IEEE802.15.4 standard, reaches a PSNR improvement of 1.95 dB, with respect to the plain transmission, while requiring an additional minimal overhead of 22.51% in the number of transmitted bits. When a higher protection level is exploited a PSNR improvement of 3.72 dB has been experienced at the cost of 40.74% additional overhead.

ScanTraffic: smart camera network for traffic information collection

Intelligent Transport Systems (ITSs) are gaining growing interest from governments and research communities because of the economic, social and environmental benefits they can provide. An open issue in this domain is the need for pervasive technologies to collect traffic-related data. In this paper we discuss the use of visual Wireless Sensor Networks (WSNs), i.e., networks of tiny smart cameras, to address this problem. We believe that smart cameras have many advantages over classic sensor motes. Nevertheless, we argue that a specific software infrastructure is needed to fully exploit them. We identify the three main services such software must provide, i.e., monitoring, remote configuration, and remote code-update, and we propose a modular architecture for them. We discuss our implementation of such architecture, called ScanTraffic, and we test its effectiveness within an ITS prototype we deployed at the Pisa International Airport. We show how ScanTraffic greatly simplifies the deployment and management of smart cameras collecting information about traffic flow and parking lot occupancy.

PyFUNS: A Python Framework for Ubiquitous Networked Sensors

In recent years Wireless Sensor Networks (WSNs) have been deployed in wide range of applications from the health and environment monitoring to building and industrial control. However, the pace of prevalence of WSN is slower than anticipated by the research community due to several reasons including required embedded systems expertise for developing and deploying WSNs; use of proprietary protocols; and limits in scalability and reliability. In this paper we propose PyFUNS (Python-based Framework for Ubiquitous Networked Sensors) to address these challenges. PyFUNS handles low level and networking functionalities, using the services provided by Contiki, and leaves to the user only the task of application development in the form of Python scripts. This approach reduces required expertise in embedded systems to develop WSN based applications. PyFUNS also uses 6LoWPAN and CoAP standard protocols to enable interoperability and ease of integration with other systems, pursuing the Internet of Things vision. Through a real implementation of PyFUNS in two constrained platforms we proved its feasibility in mote devices, as well as its performance in terms of control delay, energy consumption and network traffic in several network topologies. As it is possible with PyFUNS to easily compare performance of different deployments of distributed application, PyFUNS can be used to identify optimal design of distributed application.

Video streaming applications in wireless camera networks: A change detection based approach targeted to 6LoWPAN

Video streaming applications in wireless camera networks composed by low-end devices are attractive for enabling new pervasive high value services. When complying with the 6LoWPAN standard, the reduced amount of available bandwidth imposes the transmission of low resolution images. In this paper we present a low-complexity algorithm based on background subtraction and error resilience techniques aimed at reducing the transmission bandwidth of a video stream of uncompressed images thus permitting a higher frame rate. By means of realistic simulation studies, the performance of the presented algorithm is analyzed against state-of-the-art solutions like JPEG. These results can be considered for designing a next generation of smart cameras suited for 6LoWPAN.