In Wireless Multimedia Sensor Networks (WMSNs), maximizing network lifetime is a critical challenge due to the high energy demands of video processing and communication. Key factors influencing energy consumption include source coding parameters, the reliability and capacity of communication links, and routing strategies. Since the size of compressed multimedia data—particularly video—is inversely related to its visual quality, a trade-off naturally emerges between data volume and user-perceived quality. This paper addresses the problem of determining optimal coding and routing parameters that extend the network’s operational lifespan while maintaining the required visual quality at the sink and adapting to the dynamic nature of wireless links (e.g., fluctuating capacity and reliability). We propose a fully distributed and adaptive solution tailored to WMSNs. Our approach dynamically adjusts link utilization based on real-time reliability assessments and the residual energy of intermediate nodes. This ensures energy-efficient routing under both stable and perturbed network conditions. Extensive simulations demonstrate that our solution can prolong network lifetime by up to 9.76 times in ideal conditions and by approximately 5 times in the presence of network perturbations, while incurring only 0.16% of total battery energy as overhead. Moreover, the approach is shown to be highly responsive to topology changes, confirming its suitability for real-world WMSN deployments.

This work tackles the closed peripheral coverage issue. Such circumferential surveillance is essential in numerous realistic applications where the goal is to ensure rapid detection of any unauthorized entry/exit of the monitored area. To respond to these requirements, peripheral WMSNs can be rapidly deployed around areas of interest (e.g., temporary military camps, natural or nuclear disasters, etc.). While it seems simple, this solution raises two major challenging issues. First, checking whether the deployed peripheral WMSN forms a closed shape around the monitored region. That is, any object trying to leave or access this area must be instantaneously detected by at least one multimedia sensor node. Second, instead of activating all the deployed redundant sensor nodes, and hence, vainly wasting their scarce resources, an efficient scheduling technique must be proposed. We remedied these problems by proposing and evaluating a distributed algorithm that was specifically designed to fit the peculiarities of WMSNs.

In automotive applications, connected vehicles (CVs) can collect various information (external temperature, speed, location, etc.) and send them to a central infrastructure for exploitation in a wide range of applications: Eco-Driving, fleet management, environmental monitoring, etc. Such applications are known to generate a massive volume of data that is processed in real or near real time (i.e., data streams) depending on the target application requirements. To handle this data volume, big data architectures, based on stream computing paradigm, are usually adopted. Within this paradigm, data are continuously processed by a set of operators (elementary operations) instances. Further, a streaming application can be modeled as a directed graph where vertices are operators instances and edges are data streams (i.e., continuous series of tuples, generated by an operator). The central challenge when developing streaming applications is the way to assign operators to given resources for optimal performances (i.e., resource allocation). We showed that straightforward allocation, when based on intrinsic data parallelism, do not yield satisfying results. We developed a novel approach that takes into account the specifics of the target application, on the one hand, and on the other hand the systems performance metrics we derived from the infrastructure. The proposed approach improves the throughput by 4% compared to the previous approach. In the automotive context, this number translates into 800k additional vehicles to be served by the infrastructure over the 20M of the expected number of connected vehicles by Groupe PSA.

One of the most important issues, in Wireless Multimedia sensor networks (WMSNs), is to achieve an optimal video data encoding through an efficient resource allocation, while maximizing the overall network lifetime. However, the multimedia data is usually voluminous and delivering the compressed video data to the forwarding node poses an emergent demand, to simultaneously optimizing the video encoding and data routing performance in order to maximize the network lifetime. Based on unknown routing matrix, in this paper, we focus on the integration of routing into the analytic model proposed in our previous work. The simulation results show that our proposed solution consumes less than 0.03% of the total battery and ensures a prolongation of the network lifetime compared to the literature approaches.

Vehicles have become moving sensor platforms collecting huge volumes of data from their various embedded sensors. This data has a great value for automotive manufacturers and vehicles owners. Indeed, connected vehicles data can be used in a large broad of automotive services ranging from safety services to well-being services (e.g. fatigue detection). However, vehicle fleets send big volumes of data that traditional computing and storage approaches are not able to manage efficiently. In this paper, we present the experience of the PSA Group on leveraging big data in automotive context. We describe in depth the big data architecture deployed within the PSA Group and the underlaying technologies/products used in each component.

This paper presents a distributed serial data aggregation approach, called Spreading Aggregation~(SA), in which one packet hops sequentially over nodes and aggregates their data. The next hop of the aggregation packet is determined locally by each traversed node using only its one-hop neighborhood information, so no network topology information needs to be known by nodes, nor collisions are generated as only one node is communicating at any given time. This localized and distributed characteristic makes the proposed approach highly scalable and very efficient in terms of communication-reduction, energy conservation, and aggregation time, as confirmed by the numerous simulation results we obtained. These results confirm also the superiority of the proposed approach over the state-of-the-art serial approaches, particularly in very large scale network deployments.

The increasing usage of personal multimedia devices such as mobiles phones, smart glasses, etc. has pointed out the need for securing the captured/exchanged images content. Nevertheless, because of the limited resources of these novel platforms on one hand and the voluminous nature of multimedia content on the other hand, preserving multimedia content security remains a research challenge. In this paper, we tackle this issue by presenting a fast format-compliant selective encryption algorithm for JPEG 2000 images. It is based on selectively choosing data from the JPEG 2000 code-stream in a uniformly dynamic-key dependent manner to apply the proposed encryption algorithm. The encryption algorithm consists of two rounds of substitution-diffusion processes, based on a dynamic key, that is changed for every input image. Extensive security analysis has been conducted to evaluate the effectiveness of the proposed scheme. The obtained results have demonstrated the robustness of our algorithm against the most known types of attacks and have shown a significant improvement in term of execution time reduction compared to a similar existing JPEG 2000 images encryption scheme.

In Wireless multimedia sensor networks (WMSNs), one of the most important issues is to balance video data coding versus the perceived video quality at the sink, while maximizing the network lifetime. That is, better quality generates higher data rates and consequently consumes additional resources (i.e. energy). Finding such a balance is not obvious, especially in a distributed fashion. Previous research works have tackled this issue by not considering the routing issue; i.e. they consider that it has been done separately during the network initialization. In other term, routing is “viewed” as a network input and not as a parameter to optimize. The purpose of this paper is to demonstrate, through extensive simulations, the impact of the routing policy on the approaches used to find such a balance. To this end, we first considered several well-known routing approaches in WMSNs and extensively analyzed the impact of each one by considering three main parameters: battery consumption, activity duration, and network lifetime. The simulation results show clearly the great impact of the considered routing approaches on the obtained resource's consumption balance.

In automotive applications, several components, offering different services, can be composed in order to handle one specific task (autonomous driving for example). Nevertheless, component composition is not straightforward and is subject to the occurrence of bugs resulting from components or services incompatibilities for instance. Hence, bugs detection in component-based systems at the design level is very important, particularly, when the developed system concerns automotive applications supporting critical services. In this paper, we propose a formal approach for modeling and verifying the reliability of an autonomous vehicle system, communicating continuously with off-road infrastructure. We focus on components offering critical services with hard time constraint defining the delay of their availability. We propose to verify whether a set of components, when composed according to the system architecture specified with SysML models, achieve their tasks by respecting their interaction protocols and their time constraints.

In wireless sensor networks, a serial processing algorithm browses nodes one by one and can perform different tasks such as: creating a schedule among nodes, querying or gathering data from nodes, supplying nodes with data, etc. Apart from the fact that serial algorithms totally avoid collisions, numerous recent works have confirmed that these algorithms reduce communications and considerably save energy and time in large-dense networks. Yet, due to the path construction complexity, the proposed algorithms are not optimal and their performances can be further enhanced. To do so, in the present paper, we propose a new serial processing algorithm that, in most of the cases, approximates the optimal number of hops (i.e., it requires n - 1 communications to traverse a network of n nodes). The extensive OMNeT++ simulations confirm the outperformance and efficiency of the proposal in terms of scalability and energy/time consumption.

Wireless Multimedia Sensor Networks (WMSNs) have to deal with two main opposite constraints: (a) the voluminous nature of the sensed data, almost of megabytes order on one hand and (b) the limited resources, characterizing specifically WMSNs platforms (limited energy provision and CPU power, wireless communications, etc.) on the other hand. Additionally, as these platforms are open infrastructures, they are hence vulnerable to several types of attacks. In this paper, we propose a new and novel approach, specifically tailored to significantly reduce the transmitted multimedia data whilst allowing a high level of security, in particular data confidentiality. In fact, rather than using traditional cryptography systems which applied on multimedia data, inquiring a huge communication and hence overhead hence are not suitable within WMSNs, our approach begins first by transforming the input images using Voronoi tessellation, that reduces significantly their volume while preserving at the same time their perceived quality. As this transformation is performed on a random fashion basis, this randomness ensures then the content confidentiality. The proposed approach achieves hence both: (a) reducing noticeably the amount of the data sent by the source nodes, prolonging hence the overall network lifetime and (b) makes the network more robust against attacks. We study the parameters for setting up our approach by incorporating two schemes (basic one and more elaborated one). Finally, we demonstrate, through extensive experiments, the robustness (i.e., secure against several types of attacks) and the effectiveness of our approaches over the current state-of-the-art techniques.

Geographic greedy forwarding (GF) technique has been widely used by many algorithms for routing in sensor networks because of its high efficiency resulting from its local and memoryless nature. Hence, it ensures scalability which is a fundamental requirement for protocol applicability to large-scale sensor networks with limited resources. Nevertheless, GF suffers from a serious drawback when packets, based on geographic distance, cannot be delivered; i.e., the so-called “local minimum phenomenon”. This problem has been tackled in previous research works to guarantee packet delivery by routing around the boundaries of the hole, but at an excessive consumption of control overheads. In this paper, we propose a novel approach that exploits GF technique and guarantees at the same time packet delivery (handles the local minimum situations). Our approach is of a local nature that does not retain memories and performs better than the state-of-the-art approaches in terms of its ability to guarantee packet delivery and to derive efficient routing paths. We provide, in this paper, proof of its correctness (packet delivery guaranty) while showing, through simulations, its performance effectiveness in terms of reducing path lengths, average end-to-end delays, and overall energy consumption.

Over the last decades, research on microelectromechanical systems (MEMS) has focused on the engineering process which has led to major advances. Future challenges will consist in adding embedded intelligence to MEMS systems to obtain distributed intelligent MEMS. One intrinsic characteristic of MEMS is their ability to be mass-produced. This, however, poses scalability problems because a significant number of MEMS can be placed in a small volume. Managing this scalability requires paradigm-shifts both in hardware and software parts. Furthermore, the need for actuated synchronization, programming, communication and mobility management raises new challenges in both control and programming. Finally, MEMS are prone to faulty behaviors as they are mechanical systems and they are issued from a batch fabrication process. A new programming paradigm which can meet these challenges is therefore needed. In this article, we present CO2Dim, which stands for Coordination and Computation in Distributed Intelligent MEMS. CO2DIM is a common project between France and Hong-Kong building a new programming environment which includes a language, based on a joint development of programming and control capabilities, a simulator and real hardware.

Although Wireless Sensor Networks (WSNs) are capable of generating a huge amount of data, the ultimate objective of the underlying applications/end-users is to derive an estimate of a parameter or function of interest through queries sent to nodes containing raw data. The way these queries are handled by the network has a deep impact on its overall performances ( e.g., required communications, consumed energy, etc.). Among the numerous centralized and distributed approaches addressing this issue, serial ones have shown an interesting improvements in terms of reducing communication needed for each query and hence reducing required energy. Nevertheless, serial approaches suffer two main drawbacks: (a) they require to construct a path passing through all nodes of the network (which is known to be a NP-Complete problem) and (b) they experience poor scalability. In this paper, we investigate these issues by proposing a novel localized serial approach, called Peeling Algorithm (PA). The proposed approach, because of its localized nature (i.e., no extra-information is needed rather than what it is already available at each node), has shown better support for scalability while reducing significantly needed communications to accomplish a query. The extensive simulation evaluations we made have confirmed the effectiveness of our proposed approach in comparison to other serial approaches. We also provide in this paper formal proofs of its correctness i.e., our distributed approach terminates (free of looping) and visits all connected nodes in the network.

Besides of its high efficiency, in particular its scalability, geographic greedy routing is of a memoryless nature (i.e., nodes do not maintain any routing tables) which makes it well suitable for large-scale sensor networks with limited resources. Nevertheless, the occurrence of local minimum phenomenon, especially in networks with many holes, limits considerably the applicability of such routing protocol. The local minimum situations occur when the packet could not be delivered by the current node based only on distance to the destination. This problem has been tackled in previous research works to guarantee packets delivery by routing around the boundaries of the hole but at an excessive consumption of control overheads. In this paper, we propose and study a novel approach that is of memoryless nature and performs better that the state-of-art approaches in terms of guarantying packet delivery and deriving efficient routing paths. We provide in this paper proofs of its correctness (guarantying packet delivery) whereas showing, through simulations, its performance effectiveness in terms of reducing routing paths.

Being one of the important factors of evaluation in peer-to-peer streaming networks, availability of data is the point of our focus. Efficient data sharing in global peer-to-peer system is complicated by node failure, unreliable network connectivity and limited bandwidth. Using a caching system can improve availability of data. Yet, determining when and where to put the data in the cache to met performance goals in system with many users and files, dynamic network characteristics and changing users behavior, is difficult. We propose an approach to face these problems using a probabilistic model queue M/M/1/K and an optimization method called "Pareto method of dominance and optimality" to maximize the availability and access of data in the network. We evaluate the accuracy and performance of the proposed approach using simulations which showing the effectiveness of our approach.

This paper presents the architecture of a multimedia sensor network, especially dedicated to video surveillance. The founding ideas of our architecture are: (a) continuous 3D real-time reconstruction of the monitored area in which streams originated form video sensors are merged. (b) locating parts of data analysis/extraction tasks on the sensors themselves in a way to reduce drastically the network bandwidth usage, traditionally consumed by video stream. Hence video sensors perform in addition to capturing tasks, data analysis and extraction of features needed for the 3D reconstruction. We validated our approach through real experimentations, in particular on the video sensor.

Viewing multimedia contents through P2P streaming systems is gaining to be an emerging practice. In such systems, the steaming mode for multimedia content is used rather than the downloading mode. The advantages of this mode are numerous, especially in terms of time latency. Nonetheless, P2P streaming systems suffer from data availability problem resulting from the unpredictable peer's departure. In this paper, we present a novel approach based on the use of an embedded distributed virtual cache amongst the peers in theP2P network. An efficient allocation policy, that balances the streaming load over peers, is developed to distribute the cached data over the resulting distributed virtual cache. Additionally, we present an adaptive cache management policy, based on a closed-loop feed back control system that automatically adjusts the size of the cached data to the system workload. The simulation results we obtained show clearly the effectiveness of our approach.

Abstract In this paper, we present the design and the implementation of SIRSALE: a distributed video data management system. SIRSALE allows users to manipulate video streams stored in large distributed repositories, i.e. it provides remote users with functionalities to browse video streams by structures (shots, scenes, sequences, etc.), to annotate the semantic contents of videos and to query the distributed video repositories. One of the main contributions of SIRSALE is its contextual adaptation to the target application, i.e. it is based on a modular data model that allows adapting the system to deal with several semantic contexts. In other words, SIRSALE allows users to define and to use their own semantic data model in order to annotate and query video databases. The key idea behind this is to dynamically adapt the whole system, mainly user interfaces, to stand several semantic data models. The system has been presented to professionals who gave a positive feedback. Copyright © 2006 John Wiley & Sons, Ltd.

Multimedia data are applicable in various domains such as education, advertising, entertainment and communication. Multimedia data can be in the form of documents, and require adequate spatial and temporal presentation models. Sometimes, due to the noisy problems such as low bandwidth, or user preferences, replacing a media-element by another one can be achieved in order to provide the better quality of presentation. This paper deals with the problem of finding adequate multimedia presentation that fulfills spatio-temporal adaptation for switching between a set of alternative and semantically equivalent media elements.

We propose a buffer management policy that plans buffer sharing between multimedia presentations. Our experimental results, which are based on realistic distributions, show that the proposed approach can lead to a gain in storage bandwidth of over 15%.

This paper presents the SIRSALE system. SIRSALE is a video management system that allows users to deal with video streams stored in distributed repositories. SIRSALE is based on a modular model that allows adapting the system to deal with specific semantic contexts. Thus SIRSALE allows users to browse documents by structures (shots, scenes, sequences), to annotate the content of the video documents and to query the database by content, by using a graphical tool adapted to the user's specific topic of interest. The system has been demonstrated to professionals with a positive feedback.