Kais Loukil

Self adaptive reconfigurable system based on middleware cross layer adaptation model

The emergence of mobile multimedia systems and the diversity of the supported multimedia applications put new challenges for their design. These systems must provide a maximum application quality of service (QoS) in the presence of a dynamically varying environment (e.g. video streaming and multimedia conferencing) and multiple resources constraints (e.g. remaining energy). To respond to the changing resource availability and application demands, a new class of adaptation method is emerged. It combines the adaptation simultaneously upon the different layers related to the target system. This paper presents a framework dedicated for mobile multimedia systems. It supports application QoS under real time and lifetime constraints via coordinated adaptation in the hardware, OS, and application layer. In this framework, we present a new middleware approach based on a global and a local manager. The global manager (GM) is used to handle large and long-term variations whereas the local manager (LM) is used to guarantee the real time constraint. The GM intervenes in three layers but the LM intervenes only in the application layer and OS layer. We have implemented this approach on reconfigurable platform using Altera technology.

Self-adaptive on-chip system based on cross-layer adaptation approach

The emergence of mobile and battery operated multimedia systems and the diversity of supported applications mount new challenges in terms of design efficiency of these systems which must provide a maximum application quality of service (QoS) in the presence of a dynamically varying environment. These optimization problems cannot be entirely solved at design time and some efficiency gains can be obtained at run-time by means of self-adaptivity. In this paper, we propose a new cross-layer hardware (HW)/software (SW) adaptation solution for embedded mobile systems. It supports application QoS under real-time and lifetime constraints via coordinated adaptation in the hardware, operating system (OS), and application layers. Our method relies on an original middleware solution used on both global and local managers. The global manager (GM) handles large, long-term variations whereas the local manager (LM) is used to guarantee real-time constraints. The GM acts in three layers whereas the LM acts in application and OS layers only. The main role of GM is to select the best configuration for each application to meet the constraints of the system and respect the preferences of the user. The proposed approach has been applied to a 3D graphics application and successfully implemented on an Altera FPGA.

Energy efficient adaptive transmission strategy using cooperative diversity for wireless sensor networks

Energy saving strategies pose challenges for wireless Sensor Networks communication, as the nodes have limited energy resources and are battery powered. Thus, there is a strong interest in developing energy efficient transmission techniques. In this paper, we present a new approach to minimize the energy consumption of clustered wireless sensor networks, while achieving the required level of reliability. We consider clusters where sensors transmit monitored data to nearby cluster-head using cooperative communication. The goal of our approach is to minimize the transmit power energy based on Bit Error Probability (BEP) criterion for selective digital relaying (SDR) schemes whose closed form expression was recently derived in previous works. Therefore, the proposed scheme can efficiently allocate the least transmission power among the source and the relays that satisfy the predefined level of reliability while decreasing the energy consumption of the node. Simulation results show that our relay selection algorithm achieves significant energy savings (more than 50% compared with direct transmissions for moderate SNR values and up to 90% in high SNR regions).

A timing constraints control technique for embedded real time systems

The real-time applications have a growing complexity and size which have to be well controlled. They can be viewed as a set of synchronized tasks, communicating and sharing critical resources. One of the main difficulties in the real-time application design is time constraints meeting. All tasks have to be running before their predefined deadlines. At this level, the integration of real time operating systems (RTOS) in the real-time systems design flow is necessary to enable scheduling tasks and managing the competition between them with respect of timing constraints. One of the problems encountered here is that one task may have different execution times. It may exceed its predefined WCET and then its deadline for many reasons. The problem is that one deadline exceeding may cause subsequent constraints violations which may disrupt the functioning of the system. This paper deals with this particular issue. It presents a new technique that permits the monitoring of tasks under execution. It controls their timing constraints by means of watchdog concept and detects deadline missing. That information is used to tune the target application parameters in order to satisfy timing constraints for the further computation iterations. We have implemented this technique in the RTOS MicroC/OS-II using the EDF scheduling policy. This technique has been validated using an Altera FPGA prototyping platform and the 3D rendering application.

Optimized Parallel Model of Covariance Based Person Detection

Covariance descriptor has good performance for person detection systems. However, it has high execution time. Multiprocessors systems are usually adopted to speed up the execution of these systems. In this paper, an optimized parallel model for covariance person detection is implemented using a high-level parallelization procedure. The main characteristics of this procedure are the use of Khan Process Network (KPN) parallel programming model of computation, and the exploration of both task and data levels of parallelism. For this aim, a first KPN parallel model is proposed starting from the block diagram of the covariance person detection application. This model is implemented through the Y-Chart Application Programmers Interface (YAPI) C++ library. To ensure the best workload balance of the optimized model, communication and computation workload analysis are considered. Based on these results, both task merging and data-level partitioning are explored to derive an optimized model with the best communication and computation workload balance. The optimized parallel model obtained has three times lower execution time in comparison with the sequential model.

OS service update for an online adaptive embedded multimedia system

The increasing popularity of nomad multimedia systems put new challenges for their design: increasing functionalities, limited energy and computation resources. These systems must provide a maximum application quality of service (QoS) in the presence of a dynamically varying environment (e.g. video streaming and multimedia conferencing) and multiple resources constraints. To respond to the changing resource availability and application demands, a new class of adaptation method is emerged. It combines the adaptation simultaneously upon the dirrerent layers related to the target system: hardware, application and OS. This paper presents an overview of a multilayer. The global manager (GM) is uses to handle large and long-term variations whereas the local manager (LM) is used to guarantee the real time constraint. This paper focuses on the LM that intervenes only in the application layer and OS layer.

Images Selection and Best Descriptor Combination for Multi-shot Person Re-identification

To re-identify a person is to check if he/she has been already seen over a cameras network. Recently, re-identifying people over large public cameras networks has become a crucial task of great importance to ensure public security. The vision community has deeply studied this area of research. Most existing researches rely only on the spatial appearance information extracted from either one (single-shot) or multiple images (multi-shot) for each person. Actually, the real person re-identification framework is a multi-shot scenario. However, to efficiently model a person’s appearance and to select the most informative samples remain a challenging problem. In this work, an extensive comparison of descriptors of state of art associated to the proposed frame selection method is considered. Specifically, we evaluate the samples selection approach using different known descriptors. For fair comparisons, two standard datasets PRID 2011 and iLIDS-VID are used showing the effectiveness and advantages of the proposed method.

Classification data using outlier detection method in Wireless sensor networks

For classification data, we use Wireless sensor networks (WSNs) as hardware for collecting data from harsh environments and controlling important events in phenomena. To evaluate the quality of a sensor and its network, we use the accuracy of sensor readings as surely one of the most important measures. Therefore, for anomalous measurement, real time detection is required to guarantee the quality of data collected by these networks. In this case, the task amounts to create a useful model based on KPCA to recognize data as normal or outliers. On account of the attractive capability, KPCA-based methods have been extensively investigated, and have shown excellent performance. So, to extract relevant feature for classification and to prevent from the events, we use KPCA based on Mahalanobis kernel as a preprocessing step. In the original space, the totality of computation is done thus saving computing time. Then the classification was done on real Intel Berkeley data collecting from urban area. Compared to a standard KPCA, the results show that our method are specially designed to be used in the field of wireless sensor networks (WSNs).

Adaptive WSNs Based on HW/SW Implementation of Selective Relaying Communication Technique

Power consumption poses a challenge issue in Wireless Sensor Networks as the sensor node is power-constrained. Since Wireless communication is among the most energy expensive tasks, many power efficient transmission strategies are reported in the literature. However, most of them are purely software-based implemented. Since relaying algorithms are intensive-computing tasks, a purely software-based implementation is not the wise choice. This work comes to propose an alternative choice for implementing our earlier proposed relaying algorithm, which is a HW/SW implementation. To do so, we have identified the greedy CPU time and energy consuming tasks and implemented their dedicated hardware accelerators. We prototype our work using an FPGA board. Design details and interfacing of the different accelerators show important time execution and power consumption reduction of our algorithm.

Optimized Parallel Model of Human Detection Based on the Multi-Scale Covariance Descriptor

Human detection based on the multi-scale covariance descriptor outperforms many other antecedent descriptors. However, it has the disadvantage of being highly time consuming. The complexity of this type of application intensifies the need of multiprocessor architecture (MPSOC) to meet real time constraints. A well-balanced application model is necessary for an efficient implementation into MPSOC architecture. In this paper, a high-level independent target-architecture parallelization approach is used to propose an optimized parallel model of a multi-scale covariance human detection system. The main characteristic of this approach is the exploration of both task and data levels of parallelism. For this end, an initial model is proposed. This model is implemented and validated at a high level interface. The potential bottlenecks of this model are identified using communication and computation workload analysis. Based on this analysis, an optimized parallel model with maximum workload balance is provided. Results reveal that the obtained parallel model has more than four times lower execution time in comparison with the sequential model.