Gianluca Franchino

Energy-aware packet and task co-scheduling for embedded systems

A crucial objective in battery operated embedded systems is to work under the minimal power consumption that provides a desired level of performance. Dynamic Voltage Scaling (DVS) and Dynamic. Power Management (DPM) are typical techniques used on processors and devices to reduce the power consumption through speed variations and power switching, respectively. The effectivenes of both DVS and DPM needs to be considered in the development of a power management policy for a system that consists of both DVS-enabled and DPM-enabled components. This paper explores how to efficiently reduce the power consumption of real-time applications with constrained resources, like energy, CPU, and transmission bandwidth. A combined DVS-DP approach with a reduced complexity is proposed to make use of online strategies for embedded systems. Simulation results reveal the effectiveness of the proposed approach.

Data fusion for relative localization of wireless mobile nodes

Monitoring teams of mobile nodes is becoming crucial in a growing number of activities. When it is not possible to use fix references or external measurements, a practicable solution is to derive relative positions from local communication. In this work, we propose an anchor-free Received Signal Strength Indicator (RSSI) method aimed at small multi-robot teams. Information from Inertial Measurement Unit (IMU) mounted on the nodes and processed with a Kalman Filter are used to estimate the robot dynamics, thus increasing the quality of RSSI measurements. A Multidimensional Scaling algorithm is then used to compute the network topology from improved RSSI data provided by all nodes. A set of experiments performed on data acquired from a real scenario show the improvements over RSSI-only localization methods. With respect to previous work only an extra IMU is required, and no constraints are imposed on its placement, like with camera-based approaches. Moreover, no a-priori knowledge of the environment is required and no fixed anchor nodes are needed.

WBuST: A real-time energy-aware MAC layer protocol for wireless embedded systems

The design of wireless embedded systems for realtime applications requires a careful management of timing and energy requirements. This paper describes a wireless communication protocol that can guarantee both message deadlines and system lifetime by properly allocating the network bandwidth to each node. The protocol allows multi-hop wireless communication under different network topologies. The proposed approach is validated through both theoretical and experimental results.

Platform-aware bandwidth-oriented energy management algorithm for real-time embedded systems

A crucial objective in battery operated embedded systems is to work under the minimal power consumption that provides a desired level of performance. Dynamic Voltage and Frequency Scaling (DVFS) and Dynamic Power Management (DPM) are typical techniques used on processors and devices to reduce the power consumption through speed variations and power switching, respectively. The effectiveness of DVFS and DPM methods needs to be considered in the development of a power management policy for systems that consist of DVFS-enabled or DPM-enabled components. This paper explores how to efficiently reduce the power consumption of real-time applications with constrained resources, like energy, CPU and transmission bandwidth. A combined DVFS-DPM approach with a reduced complexity is proposed to make use of on-line strategies for embedded systems.

BACCARAT: a dynamic real-time bandwidth allocation policy for IEEE 802.15.4

Recently, researchers and engineers began considering the use of WSN in time-sensitive applications. For effective real-time communications, it is important to solve the problem of contention to the communication medium providing an efficient bandwidth allocation mechanism. In this paper we tackle with the problem of performing timely detection of events by a WSN. We propose a real-time bandwidth allocation mechanism for IEEE 802.15.4 that maximizes event detection efficiency and reduces statistical uncertainty under network overload conditions. On-line strategies complement off-line guarantees to enhance the confidence level of the measurements.

BuST: Budget Sharing Token protocol for hard real-time communication

Timed-token networks, such as FDDI, support both synchronous real-time traffic and non real-time traffic (asynchronous messages). The medium access scheme of FDDI guarantees up to one half of the total network bandwidth for synchronous communication. Further enhancements, such as FDDI-M, improve the bandwidth dedicated to real-time messages. However, the ability of timed-token protocols to guarantee synchronous message deadlines highly depends on specific Synchronous Budget Allocation (SBA) schemes. This paper introduces BuST, the Budget Sharing Token protocol which improves the management of periodic real-time traffic, while guaranteeing a minimum throughput for non real-time messages, with respect to existing techniques. We evaluate the performance of BuST, in comparison with FDDI and FDDI-M, considering a Synchronous Budget Allocation (SBA) scheme proposed in the literature, using the Worst-Case Achievable Utilization (WCAU) as performance metrics. We demonstrate that the performance achieved by BuST is better or, in a few cases, equal to FDDI and FDDI-M.

Token Passing Techniques for Hard Real-Time Communication

Distributed computing platforms are increasingly used to develop critical embedded systems, like control applications, sensors networks, telecommunication, and robotics systems. In such distributed applications, the correct behaviour depends on the timely execution of the tasks running on different nodes, which may frequently exchange shared data. In particular, since the nodes are typically connected through a common channel without the need of multi-hop communication, it turns out that a timely communication mainly depends on how the nodes access the channel. Even when a multi-hop communication is needed, a timely message delivery is not feasible without the support of a predictable channel access mechanism, which is implemented in the Medium Access Control (MAC) sub-layer of the communication stack. There exist several MAC protocols designed for providing a timely communication among distributed nodes, mainly in the factory communication domain. One of the most effective solutions is given by token passing protocols, which have some nice characteristics that make them suitable for real applications. For instance, because of the token passing mechanism, the nodes do not need to be synchronized and they have an implicit bandwidth reclaiming mechanism that allows other nodes to exploit the unused bandwidth. Moreover, such protocols can serve both real-time and best-effort (non real-time) traffic. Among token passing protocols, the timed token policy is a channel scheduling approach first proposed by Grow in (Grow, 1982). Since then, it has received a substantial attention and several relevant results have been derived (Zhang et al., 2004), which make timed token protocols suitable for the real-time communication in industrial applications. Timed token policies are used as channel access mechanism in several standard protocols such as, for instance, PROFIBUS (Profibus, 1996) and FDDI (FDDI, 1987). However, the application domain of the timed token policy is not restricted to the cited communication standards, and some examples on their use can be found in (Lenzini et al., 2004) and (Cicconelli et al., 2007). To improve the ability of timed token protocols of managing real-time traffic, Shin and Zheng (Shin & Zheng, 1995) proposed a modification of the timed token protocol, which can guarantee a greater bandwidth for the real-time traffic with respect to the classic timed token protocol; however, under certain conditions, it cannot manage the best-effort traffic (Franchino et al., 2008). The Budget Sharing Token (BuST) protocol has been proposed as an improvement with respect to existing timed token protocols (Franchino et al., 2008). In 12

A distributed architecture for mobile robots coordination

In many application fields, such as surveillance, monitoring, exploration, and rescuing, the use of multiple coordinated robot units seems to be the most convenient solution, in terms of costs, performance, and efficiency. In this paper we present a distributed real-time architecture for coordinating a set of mobile robot that have to operate according to a common objective. The interaction with the environment causes each unit to operate under real-time constraints, which are enforced by the operating system to achieve a desired level of performance. The design of the architecture has been focused on three main aspects: real-time communication, modularity, and flexibility. The last two aspects are essential in a distributed environment to simplify maintenance and system upgrades. The proposed approach has been implemented on a team of six robot vehicles, equipped with proximity and light sensors that can operate according to predefined formations

A power-aware MAC layer protocol for real-time communication in wireless embedded systems

The number of battery-operated devices with wireless communication capabilities has increased enormously in the last years and is expected to increase even more in the future. A fundamental need in these systems is to guarantee a minimum system lifetime and timing constraints through a careful management of energy, communication, and computational resources. This paper describes WBuST, a MAC layer protocol designed to bound the maximum delay of real-time messages and guarantee the system lifetime by properly allocating a share of the available bandwidth to each node of the network. The protocol allows multi-hop wireless communication under different network topologies. The proposed approach is assessed through theoretical analysis and experimental results.

Energy-aware algorithms for tasks and bandwidth co-allocation under real-time and redundancy constraints

The energy consumption in distributed systems depends on several inter-related factors, including task partitioning, process redundancy, fault tolerance, task and message scheduling, and communication bandwidth allocation. Although some of these issues have been considered in the literature in isolation, a systematic approach considering all the constraints is still missing. This paper addresses the problem of allocating a task set and the required communication bandwidth on a distributed embedded system, aiming at reducing energy consumption while guaranteeing timing and redundancy constraints. Two heuristic approaches are proposed and compared against a complete method and simulated annealing. Simulation results show the effectiveness of the proposed approaches.