QUAlity of ServIce for wireless sensor network and Mobile Objects – parameter aDaptatiOn – QUASIMODO

For efficiently tackling those issues, methods combining control theory, optimization and networking approaches are developed. The main technology that has been extensively investigated is the wireless sensor networks.
The first issue is addressed by including delay and packet loss in the extended Kalman filters for multi-target tracking, and by developing robots (mobile sensor nodes) motion algorithms optimizing moving distance. MAC-routing cross-layer approach is applied to deal with the second issue.

The outcomes include innovative protocols (MAC and routing) capable of providing the best state-of-the-art QoS and energy optimization in WSN as well as new algorithms for high efficient multi-target tracking.
For supporting real-time multi-target tracking, we addressed several fundamental issues. Barrier coverage problem has been addressed using curve-based optimization and optimal sensor deployment is realized by developing two new algorithms. For saving energy, active sensor scheduling using ultrasonic sensor networks has been addressed. We proposed a distributed saturation degree based algorithm that jointly schedules the TDMA slots for channel access and the sensor activations for avoiding the inter-sensor interference. We showed that this algorithm is scalable and energy efficient. Mobile target tracking with guaranteed coverage is another important issue. We developed two algorithms for both localization taking into account the target size and hidden target problem, and mobile actuator motion control which minimizes the energy under the sensor coverage and radio connectivity constraints.
For providing QoS in WSN, a hybrid CSMA-TDMA MAC protocol called iQueue-MAC has been designed and implemented on STM32W108 chips. It self-adapts to the traffic variation and more efficient than the existing ones to carry burst traffic (real-time QoS), while still preserving power consumption. For multi-hop routing under multi-constraints (energy, link reliability, delay), we developed a routing protocol based on the operator calculus algebra. It has been successfully implemented on Contiki/TelosB platform. Finally an important progress has been made on the end-to-end delay estimation in multi-hop WSN with dynamic MAC and routing protocol. We introduced a new approach that uses the code execution instruction for building the Markov chain model of the node behavior. This allows solving the actual difficulty to establish analytic performance models for highly dynamic networks.

From scientific point of view, this project allowed to producing numerous high-ranking publications (15 journal papers and 16 conferences papers) on both WSN protocols and target tracking using WSN. We can also notice that the achieved results represent a spectrum larger then the initial project objectives, which is normal taking into account the rapid evolution of the domain.
Several results have made progress the state of the art. For instance, iQueue-MAC is the best-known one in terms of its self-adaptability to the traffic variation while still keeping low power during light traffic period. This is why we continue to push it by aiming at making its source code available to the community on RIOT OS platform, pushing it to be adopted as one of the duty-cycle MAC protocols for the future IoT applications.

In total 15 journal papers and 16 conferences papers have been published including in high-ranking journals and conferences.
Below are some of the selected ones:
[1] Xiufang Shi , Ye-Qiong Song, Zaiyue Yang and Jiming Chen, «Multiple Target Tracking under Occlusions Using Modified Joint Probabilistic Data Association«, Proceedings of IEEE ICC2015 (accepted)
[6] Shuguo Zhuo, Zhi Wang, Ye-Qiong Song, Zhibo. Wang, and Luis Almeda, iQueue-MAC: A traffic adaptive duty-cycled MAC protocol with dynamic slot allocation, Proceedings of IEEE SECON, 2013.
[10] Jing Bai, Peng Cheng, Jiming Chen, Guenard, A., Yeqiong Song, Target Tracking with Limited Sensing Range in Autonomous Mobile Sensor Networks, Proceedings of DCOSS, 2012.
[11] Shuguo Zhuo, Ye-Qiong Song, Zhi Wang, Zhibo Wang, Queue-MAC: A queue-length aware hybrid CSMA/TDMA MAC protocol for providing dynamic adaptation to traffic and duty-cycle variation in wireless sensor networks, Proceedings of WFCS, 2012.
[18] Shibo He, Xiaowen Gong, Junshan Zhang, Jiming Chen, Youxian Sun, Curve-Based Deployment for Barrier Coverage in Wireless Sensor Networks, IEEE Transactions on Wireless Communications, 13(2): 724-735, Feb, 2014.
[24] Peng Cheng, Fan Zhang, Jiming Chen, Youxian Sun and Xuemin Shen. A Distributed TDMA Scheduling Algorithm for Target Tracking in Ultrasonic Sensor Networks, IEEE Transactions on Industrial Electronics, 60(9): 3836-3845, Sept, 2013.
In addition, two Quasimodo invited sessions have been organized. One at the IEEE WiSARN Spring Workshop in conjunction with IEEE DCOSS’2012 conference, and another at the 1st International Workshop on Compressive Sensing in Cyber-Physical Systems (CSCPS) in conjunction with IEEE MASS 2013. Partial Quasimodo project results have also been presented at LIAMA France-Chine50 workshop in Paris in May 2014.

The main scientific and technical objective of QUASIMODO is to specify, develop and evaluate mechanisms for providing the adaptive QoS in WSAN that supports real-time applications and therefore to demonstrate the effective use of WSAN technology for building CPS (Cyber-Physical Systems).
The scientific and technical challenges are:
- At the application level: there are not yet communication models and properties as a safe support for the cooperation and coordination within a system composed of mobile objects and WSN. In fact, consider rich sensory data in WSAN scenario, traditional MRS methodologies and results may not be satisfactory. Impact of WSAN performance (especially data delivery rate and delay, network nodes’ lifetime and node mobility) on the application level actions (e.g. mobile objects tracking precision, cooperation and coordination quality) is still unclear and must be carefully investigated within this project. The suitable communication model must then be specified and needed QoS ensured (e.g. temporal and space consistency of the data from distributed sensors should be ensured for correct robots cooperation and coordination).
- At the network level, providing QoS in WSAN is itself an open research topic since traditional wired network QoS parameters must be reconsidered taking into account the WSAN specificity (power and resource constraint, data-centric paradigm, node mobility, etc.). Subtle trade-offs must be found between energy consumption, the degree of redundancy and real-time constraints.
- Most importantly, at application and network interface level, there is little result on how to map the application requirements on to the network QoS parameters, and how the application can take into also account the specific characteristics of the WSAN, all these in an on-line interactive way. For this purpose more investigation on the middleware development has to be carried out for providing adaptive QoS scheme: dynamic network resource allocation to adapt to the application requirements, and the application performance adaptation to the network performance changes.

As a consequence, expected scientific advances are:
(a) For monitoring and control applications using WSAN with mobile robots, a set of well specified communication models/schemes for basic functions like tracking, coordination and cooperation, together with the variation of each of them in function of network QoS (delay, reliability, energy).
(b) A set of innovative network condition aware QoS mechanisms including energy efficient QoS routing and MAC protocols,Routing-MAC cross-layer optimisation dynamic network topology management (thanks to the mobile nodes).
(c) An easy “application-customerizable” middleware providing QoS adaptation between network and application.

In summary, providing the adaptive QoS in WSAN and demonstrating the generality of the proposed architecture is ambitious objective. We planned to demonstrate its achievement through intensive simulations and experimental development on a well-specified test bed. Moreover, as a side product, the simulator that we will develop could become an open GL software tool for further use to help the WSAN-based application designer to configure and optimize his system.