Search

Signals from gyros and accelerometers in an inertial navigation system (INS) can be processed to obtain navigation information while its errors accumulate with time. A path with long‐term characteristics is more likely to encounter unpredictable dynamic conditions. To deal with such problems, this work improves the global static path‐planning algorithm based on the A* algorithm, and the local dynamic path‐planning algorithm based on dynamic window approach (DWA). In the global planning algorithm, the gravity index and the heading index are incorporated into the cost function of the A* algorithm to consider the mismatches and the deflection of the path. Furthermore, the search method of the A* algorithm is also improved. In the local planning algorithm, the motion model of the DWA algorithm is improved to a high‐precision model and expressed in a geographic coordinate system. Finally, the improved global static planning and local dynamic planning are combined as the global dynamic path‐planning algorithm of an underwater vehicle. The simulation results show that the global dynamic path‐planning algorithm proposed can realise obstacle‐free navigation of the underwater vehicle.

Transactional memory has been attracting increasing attention in recent years, and it provides optimistic concurrency control schemes for shared-memory parallel programs. The rapid development and wide adoption of transactional memory make this programming paradigm promising for achieving breakthroughs in massively parallel computing. There has been a large number of discussions towards transactional memory systems, which aimed at providing relatively simple and intuitive synchronization construction for shared-memory parallel programs without sacrificing performance. Hardware transactional memory (HTM) has become commercially available in mainstream processors, however, due to several inherent architectural limitations that will abort hardware transactions, such as cache overflows, context switches, hardware as well as software exceptions, etc., nowadays HTM systems come in a best-effort way, which necessitates the adoption of a software fallback path to ensure forward progress. In this paper, we survey state-of-the-art software-side optimizations for best-effort hardware transaction system, as well as several novel performance tuning techniques. Research efforts about joint usage of HTM and non-volatile memory (NVM) are also discussed.

Great progress has been made in the field of pedestrian detection, but the following two problems have not yet been well addressed. One problem is the missed detection of small scale pedestrain as false negative failure, and the other one is confusion with anthropomorphic negative samples like vertical structures as false positive failure. In this paper, to tackle the above two problems, we use the light-field camera to capture pedestrian images for the following reasons: (i) the light-field camera can obtain multi-depth refocused images in a single exposure using one sensor, (ii) compared with 2D images, these refocused images can provide different key representations for different parts of the image. We further establish a light-field pedestrian dataset with 1766 images for pedestrian detection. A multi-focus detection network proposed in this work consists of multiple-branch detection models and takes multiple refocused images as inputs. In order to select the appropriate candidate proposal bounding box as final detection results, we design a cumulative probability selection (CPS) layer to combine each refocused image branch and accumulate the probability of each candidate neighboring proposal. Experimental results demonstrate that the proposed method outperforms state-of-the-art methods on our light-field pedestrian dataset.

The positioning accuracy of gravity matching is closely related to the distribution of gravity anomaly in matching area. Matching area selection is the key to improve the gravity matching accuracy. In this paper, a novel analysis method for matching area selection is proposed to solve the problem that the existing methods show strong subjectivity and ignore the directions. Firstly, the existing gravity characteristic parameters were selected by correlation analysis. Then a basic adaptability factor was built by combining the selected parameters through the subjective and objective weighting method. Finally direction factors were added to build comprehensive characteristic parameters in four main directions through multiple linear regression. Simulation result shows that the average matching rate is larger than 0.7 in the matching area selected by comprehensive characteristic parameters and the matching rate rank of different directions is consistent with comprehensive characteristic parameters rank in local area. The proposed comprehensive characteristic parameters can effectively value the adaptability of matching area in different directions.

In this paper, we propose a real-time gesture recognition model based on a feedforward artificial neural network for surface electromyography (sEMG) signals, which contains four processes: pre-processing, feature extraction, classification, and post-processing. In the feature extraction and selection stage, sliding windows are used to segment the data, and time-domain features and convolution results are used as input features. To avoid over-fitting of training, a dropout layer is added to the neural network model. In the post-processing part, two statistical methods are used: majority voting and continuous voting result verification. Finally, a test trial was performed using sEMG signals collected from the human body. The results show that the algorithm can achieve 92.7% recognition accuracy and has good real-time performance.

Consider the problems of low accuracy and poor real-time performance in traditional ICCP gravity matching algorithm, this paper proposes an improved ICCP gravity matching algorithm based on Mahalanobis distance. Mahalanobis distance, instead of Euclidean distance, is applied to improve the objective function of rigid transformation iteration in traditional ICCP algorithm, which can eliminate the interference of variable correlation. At the same time, a new searching strategy based on strip search area is proposed. With the current sampling point as the center, we use the gravimeter measurement error variance to determine the width of the strip region along the normal direction of the contour, and use the Mahalanobis distance objective function value obtained in the iterative process as the length of the strip area along the direction of the contour. Therefore, this searching strategy can gradually narrow the search range of the closest contour point in the ICCP algorithm, and reduce the calculation amount of the algorithm. The simulation results show that: Compared with the traditional ICCP algorithm, the mean value of the position error of the improved algorithm is reduced by 26.89%, and the standard deviation is reduced by 58.53%, which greatly improves the accuracy of the gravity matching algorithm. And the average running time of the improved algorithm is also 42.1% shorter than the traditional ICCP algorithm, so the real-time performance is improved. The improved ICCP gravity matching algorithm based on Mahalanobis distance reduces the running time of the algorithm, improves real-time performance and positioning accuracy, and further improves the practicability of gravity matching for underwater vehicles.

Geometric curves are mostly applied as guidance for lane-keeping and lane-changing motions. In this work, we explore the capabilities of using sigmoid functions in vehicle overtaking scenarios. Such curves are parameterized based on the current relative distance and speed concerning the preceding vehicle. Besides, the overtaking decision, i.e., whether to change lane or not, can be determined by using a piecewise function in the planning module. Then, model predictive control (MPC) is applied to track the generated overtaking path. Case studies including two scenarios, i.e., preceding vehicle driving at constant and varying speed, show that the proposed control framework is capable of self-decision making and handling overtaking maneuver effectively.

In recent years, the application of automated guided vehicles (AGVs) in industrial manufacture has attracted more and more attention. Based on reinforcement learning, an automated guided logistics handling vehicles routing algorithm was proposed under a multi-task scenario. First, the algorithm calculated routing paths for each task node using Q-learning to solve the “curse of dimensionality” problem. Then, with the objective of comprehensive cost considering the time window, the optimal routing sequence was obtained using the depth first search algorithm (DFS). Finally, an example is given to illustrate the algorithm.

Electric Vehicle (EV) sharing systems have recently experienced unprecedented growth across the world. One of the key challenges in their operation is vehicle rebalancing, i.e., repositioning the EVs across stations to better satisfy future user demand. This is particularly challenging in the shared EV context, because i) the range of EVs is limited while charging time is substantial, which constrains the rebalancing options; and ii) as a new mobility trend, most of the current EV sharing systems are still continuously expanding their station networks, i.e., the targets for rebalancing can change over time. To tackle these challenges, in this paper we model the rebalancing task as a Multi-Agent Reinforcement Learning (MARL) problem, which directly takes the range and charging properties of the EVs into account. We propose a novel approach of policy optimization with action cascading, which isolates the non-stationarity locally, and use two connected networks to solve the formulated MARL. We evaluate the proposed approach using a simulator calibrated with 1-year operation data from a real EV sharing system. Results show that our approach significantly outperforms the state-of-the-art, offering up to 14% gain in order satisfied rate and 12% increase in net revenue.

Byte-addressable non-volatile memory (NVM) makes it possible to perform fast in-memory accesses to persistent data using standard load/store processor instructions. Some approaches for NVM are based on durable memory transactions and provide a persistent programming paradigm. However, they cannot be applied to existing multi-threaded applications without extensive source code modifications. Durable transactions typically rely on logging to enforce failure-atomic commits that include additional writes to NVM and considerable ordering overheads.This paper presents PMThreads, a novel user-space runtime that provides transparent failure-atomicity for lock-based parallel programs. A shadow DRAM page is used to buffer application writes for efficient propagation to a dual-copy NVM persistent storage framework during a global quiescent state. In this state, the working NVM copy and the crash-consistent copy of each page are atomically updated, and their roles are switched. A global quiescent state is entered at timed intervals by intercepting pthread lock acquire and release operations to ensure that no thread holds a lock to persistent data.Running on a dual-socket system with 20 cores, we show that PMThreads substantially outperforms the state-of-the-art Atlas, Mnemosyne and NVthreads systems for lock-based benchmarks (Phoenix, PARSEC benchmarks, and microbenchmark stress tests). Using Memcached, we also investigate the scalability of PMThreads and the effect of different time intervals for the quiescent state.

Non-volatile random-access memory (NVRAM) technology is maturing rapidly and its byte-persistence feature allows the design of new and efficient fault tolerance mechanisms. In this paper we propose the versionized process (VerP), a new process model based on NVRAM that is natively non-volatile and fault tolerant. We introduce an intermediate software layer that allows us to run a process directly on NVRAM and to put all the process states into NVRAM, and then propose a mechanism to versionize all the process data. Each piece of the process data is given a special version number, which increases with the modification of that piece of data. The version number can effectively help us trace the modification of any data and recover it to a consistent state after a system crash. Compared with traditional checkpoint methods, our work can achieve fine-grained fault tolerance at very little cost.

New non-volatile memory (e.g., phase-change memory) provides fast access, large capacity, byte-addressability, and non-volatility features. These features, fast-byte-persistency, will bring new opportunities to fault tolerance. We propose a fine-grained checkpoint based on non-volatile memory. We extend the current virtual memory manager to manage non-volatile memory, and design a persistent heap with support for fast allocation and checkpointing of persistent objects. To achieve a fine-grained checkpoint, we scatter objects across virtual pages and rely on hardware page-protection to monitor the modifications. In our system, two objects in different virtual pages may reside on the same physical page. Modifying one object would not interfere with the other object. This allows us to monitor and checkpoint objects smaller than 4096 bytes in a fine-grained way. Compared with previous page-grained based checkpoint mechanisms, our new checkpoint method can greatly reduce the data copied at checkpoint time and better leverage the limited bandwidth of non-volatile memory.

Real-time collaborative programming is an emerging approach that supports a team of programmers to view and edit shared source code at the same time. Each programmer can edit any part of the source code, and changes become instantly visible at other collaborating sites. With a broad range of application scenarios and benefits, real-time collaborative programming has attracted increasing interest from academia and industry. Lightweight integrated development environments (lightweight IDEs) have rapidly grown in popularity in the recent years, but there are serious problems and limitations with existing real-time collaboration support for lightweight IDEs. In this study, we contribute a novel real-time collaborative programming environment named CoVSCode that supports unconstrained and flexible real-time collaboration based on Visual Studio Code, one of the most popular and widely used lightweight IDEs. We present design objectives and rationales, the workflow and functional design from collaborating programmers&rsquo, perspectives, major technical issues and solutions, and prototype implementation, as well as a set of experimental evaluations that have demonstrated the technical feasibility and good performance of the prototype system. All approaches, techniques and solutions derived in this work are generic, which can also be applied in building real-time collaborative programming environments for other lightweight IDEs.

QuiescentNVM is a user-space runtime providing transparent failure-consistency guarantees for lock-based parallel programs executing on hybrid combinations of traditional DRAM and byte-addressable non-volatile memory (NVM) technologies. A dual-versioning mechanism performs in-place persistent writes over one copy, and consistent fallback guarantees are provided by the other copy. Thus, the two writes to NVM present in logging-based solutions (such as for durable memory transactions) are reduced to a single write. Further, we avoid the need to rewrite legacy applications to exploit durable transactions. Our system relies on its dual-copy framework operation that safely persists data during global quiescent states, where no thread must hold a lock on persistent data. For applications with low lock-contention, global lock-free quiescent states will occur sufficiently frequently, and we deliver better performance and lower wear to NVM than current systems. We do not cover high-lock contention scenarios while enforcing quiescent states to occur.

In gravity aided inertial navigation, a good path planning, which makes the path pass the area where the gravity changes drastically, in matching area will improve the matching accuracy effectively. The underwater vehicle is restricted by kinematics and other constraints, and cannot move in any direction. However, The path planned by the traditional path planning algorithm is a broken line and is not suitable for actual underwater vehicle navigation. Aiming at underwater vehicle path planning problem, this paper proposes a gravity aided inertial navigation path planning algorithm which is based on vehicle constraints and Bezier curves. The $A^{\ast}$ algorithm is optimized by vehicle constraints and gravity characteristics. The Bezier curve is used to smooth the planned path, making the planned path more suitable for the actual situation. Simulation experiments show that the matching accuracy is effectively improved by the planned path based on vehicle constraints and Bezier curve.

Based on the historical load data collected in July and August 2020 in Chongqing, and in view of the characteristics of the regional load in summer Chongqing, a new short-term load forecasting method was put forward in this paper, which is based on BP neural network and takes into account social activities and meteorological factors. The simulation results show that the proposed short-term load forecasting method can accurately realize the daily maximum load forecasting, it provides a theoretical reference for load forecasting and power resource optimization of smart grid in the future.

Unlike terrestrial networks that mainly rely on radio waves for communications, underwater networks utilize acoustic waves, which have comparatively lower loss and longer range in underwater environments. However, acoustic waves incurs long propagation delays that typically lead to low throughput and higher energy cost of transmission than reception. Thus, collision and retransmission should be reduced in order to reduce energy cost and improve throughput. Receiver-based protocols, like receiver-initiated packet train and cluster-based on-demand time sharing, can significantly reduce collision and retransmission. But they are either time-consuming or energy consuming because nodes are controlled to turn into receiver mode by control packets or a timer regardless of load. In this paper, we propose an underwater practical MAC protocol (UPMAC), whose objective is to be adaptive to the network load conditions by providing two modes (high and low load modes) and switching between them based on different offered load. Turn-around time overhead is reduced and it is less vulnerable to control packet corruption by using the technique of piggyback. UPMAC uses receiver-based approach in high load mode, leading to a low data collision rate in both one-hop and multihop situation. Extensive simulations show that UPMAC can achieve better performance in both general and Sea Swarm topologies.

In this paper, a network illegal access detection method based on PSO-SVM algorithm in the power monitoring system is proposed, where the particle swarm optimization (PSO) algorithm is used to optimize the parameters of support vector machine (SVM). And the proposed method is used to classify normal data flow and abnormal data flow in the network to detect whether there is illegal access behavior in the power monitoring system. Compared with the original SVM-based method and LS-SVM-based method, the proposed method not only improves the convergence speed of the network training, but also improves the accuracy rate. The proposed method can fleetly discovery and locate the illegal access behavior in the power monitoring system, which is a meaningful study in improving the security of the power monitoring system.

Modern computer systems are suffering from various failures and are increasingly prone to error. This paper introduces a high efficient incremental checkpoint mechanism for fault tolerance. By leveraging the new hardware support for various page sizes, we make dynamic transformation between different sized pages and thus achieve the best balance of overhead between modification tracking and data dumping in incremental checkpoint. Experiments show that our new checkpoint mechanism could achieve an average speedup of 2.2x with the space overhead of 34% over traditional incremental checkpoint, showing great potential to be widely adopted.

Page protection is often used to achieve memory access monitoring in many applications, dealing with program-analysis, checkpoint-based failure recovery, and garbage collection in managed runtime systems. Typically, low overhead access monitoring is limited by the relatively large page-level granularity of memory management unit hardware support for virtual memory protection. In this paper, we improve upon traditional page-level mechanisms by additionally using hardware support for virtualization in order to achieve fine and flexible granularities that can be smaller than a page. We first introduce a memory allocator based on page protection that can achieve fine-grained monitoring. Second, we explain how virtualization hardware support can be used to achieve dynamic adjustment of the monitoring granularity. In all, we propose a process-level virtual machine to achieve dynamic and fine-grained monitoring. Any application can run on our process-level virtual machine without modification. Experimental results for an incremental checkpoint tool provide a use-case to demonstrate our work. Comparing with traditional page-based checkpoint, our work can effectively reduce the amount of checkpoint data and improve performance.

With shrinking design rules, the overall patterning requirements are getting aggressively tighter and tighter. For the 5-nm node and beyond, on-product overlay below 2.5nm is required. Achieving such performance levels will not only need optimization of scanner performance but a holistic tuning of all process steps. In previous work, it has been shown that process-induced pattern asymmetry has significant impact on overlay performance at wafer edge and can be partially compensated by applying high-order scanner corrections or optimizing metrology targets. Today, we present the reduction of process-induced pattern asymmetry in a tunable etch system and demonstrate the related on-product overlay improvement combined with scanner corrections. In our work we utilize etch tools (Lam Kiyo® conductor etch systems) with proprietary edge tuning technology that can be used to reduce the etch-related asymmetry at the wafer edge. In combination to this unique method, we evaluate the impact of high order corrections per exposure field to compensate for process asymmetry at the wafer edge with a state-of-the-art 1.35 NA immersion scanner (NXT:1970Ci). The study is done on dedicated test wafers with 10-nm logic node design. We use angle-resolved scatterometry (YieldStar® S-250), atomic force microscopy, and SEM cross-sections to characterize process asymmetry. We present experimental investigation of the effect of etch tuning and scanner corrections on the pattern shift and the resulting overlay. In particular, we present results showing a reduction of etch-induced pattern shift by 12nm at wafer radius 147mm. Results show that asymmetry can be addressed by both, litho compensation and etch tuning, and bring on-product overlay down to the required level. We discuss the benefit of the correction techniques especially for thick hard mask layers (the pattern shift scales linear with hard mask thickness) and evaluate a combined correction scenario, where preventive etch tuning and feed-back based scanner corrections are used. We conclude that a holistic tuning of all process steps will be required to fulfill overlay requirements of future nodes.

SUMMARY The nondeterminism of message-passing communication brings challenges to program debugging, testing and fault-tolerance. This paper proposes a novel deterministic message-passing implementation (DMPI) for parallel programs in the distributed environment. DMPI is compatible with the standard MPI in user interface, and it guarantees the reproducibility of message with high performance. The basic idea of DMPI is to use logical time to solve message races and control asynchronous transmissions, and thus we could eliminate the nondeterministic behaviors of the existing message-passing mechanism. We apply a buffering strategy to alleviate the performance slowdown caused by mismatch of logical time and physical time. To avoid deadlocks introduced by deterministic mechanisms, we also integrate DMPI with a lightweight deadlock checker to dynamically detect and solve these deadlocks. We have implemented DMPI and evaluated it using NPB benchmarks. The results show that DMPI could guarantee determinism with incurring modest runtime overhead (14% on

Web page segmentation aims to break a page into sections that can reveal the information presentation structure and appear coherent to readers. In this paper, we propose a new web page segmentation framework based on the process of analyzing and understanding web page structure. After extracting the segmentation graph structure, we formulate the label assignment task which determines whether each boundary should segment current block or not on a graph as a structured learning problem. Computation of highest scoring label assignment relies on Viterbi algorithm and joint feature function captures the dependency among boundaries. To solve the learning of parameters, we adopt a learning model based on perceptron algorithm. Furthermore, utilizing the previous framework, we propose a web information crawling application framework which integrates web page segmentation and semantic block classification process.

Currently, various Medium Access Control (MAC) protocols have been proposed for underwater Wireless sensor networks. Unlike terrestrial networks, underwater networks utilize acoustic waves, which have comparatively lower loss and longer range in underwater environments. However, the use of acoustic waves incurs long propagation delays that typically lead to low throughput especially in protocols that require receiver feedback such as multimedia stream delivery and the energy cost of transmission is much higher than receptions. Thus, collision and retransmission should be reduced in practice in order to reduce energy cost and improve throughput. Based on these motivations, we propose a novel MAC protocol called NCDC-MAC. NCDC-MAC leverages network coding and duty cycle, the combination of which is seldom explored, to solve these challenges. Heterogeneous wireless networks and node roles are considered while designing our algorithms. Meanwhile, fairness including schedule and service time assignment is supported in our approach. Extensive simulations show that our approach can achieve significantly better performance.

Huge pages have been widely supported by architecture and operating system. Huge pages map large fixed virtual memory regions, on the orders of 2MB to 1GB on Intel x86-64 architecture. The page size is the key to striking the balance between trade-off pairs. For example, initially, the use of huge pages aims to mitigate address translation overhead for memory-intensive workloads with large memory footprints. The huge pages do benefit to the TLB hit rate as well as the TLB miss penalty but occupy a huge amount of contiguous physical memory. Moreover, the use of huge pages influences the overhead of page-based technologies. For example, the incremental checkpoint keeps track of memory access information at the granularity of the memory page. The page size is the basic unit ofthe range of monitoring. Itdetermines the frequency of page faultsand the copying overhead, which are another trade-off pair. This paper establishes the Dynamic Page Size Adjustment (DPSA) framework in Linux OS. This work provides merge and split operations in OS that change the size of memory pages during runtime. We also support on and off DPSA modes to enable an automaticadjustment of the page size based on memory access pattern. These functions are all exposed via system-level APIs. This framework is free to use for application programmers. They can either turn off the mode tocall the APIsaccording to their applications' characteristic, or they can turn on the mode to give freedom to the DPSA.

The Mean Time to Failure continues to decrease as the scaling of computing systems. To maintain the reliability of computing systems, checkpoint has to be taken more frequently. Incremental checkpointing is a well-researched technique that makes frequent checkpointing possible. Fine-grained incremental checkpointing minimizes checkpoint size but suffers from significant monitoring overhead. We observe the memory access at page granularity and find that the size of contiguous memory regions visited by applications tends to be proportional to size of corresponding memory allocation. In this paper, we propose the Application-Based Coarse-Grained Incremental Checkpointing (ACCK) that leverages the priori information of the memory allocation to release the memory monitoring granularity in an incremental and appropriate way. This provides better opportunities for balancing the tradeoff between monitoring and copying overhead. ACCK is also assisted by hugepage to alleviate the TLB overhead. Our experiment shows that ACCK presents 2.56x performance improvement over the baseline mechanism.

Free-space optical (FSO) links offer gigabit per second data rates and low system complexity, but suffer from atmospheric loss due to fog and scintillation. Radio-frequency (RF) links have lower data rates, but are relatively insensitive to weather. Hybrid FSO/RF links combine the advantages of both links. Currently, selection or "hard-switching" is performed between FSO or RF links depending on feedback from the receiver. This technique is inefficient since only one medium is used at a time. In this paper, we develop a "soft-switching" scheme for hybrid FSO/RF links using short-length Raptor codes. Raptor encoded packets are sent simultaneously on both links and the code adapts to the conditions on either link with very limited feedback. A set of short-length Raptor codes (k = 16 to 1024) are presented which are amenable to highspeed implementation. A practical Raptor encoder and decoder are implemented in an FPGA and shown to support a 714 Mbps data rate with a 97 mW power consumption and 26360 gate circuit scale. The performance of the switching algorithms is simulated in a realistic channel model based on climate data. For a 1 Gbps FSO link combined with a 96 Mbps WiMAX RF link, an average rate of over 472 Mbps is achieved using the implemented Raptor code while hard-switching techniques achieved 112 Mbps on average.

Wireless sensor networks may be destroyed by targets in field surveillance. Self-protection technology focuses on using sensor nodes to protect themselves so as to resist attacks. Selecting protecting nodes has been proved to be a NP-complete problem. In this paper, we propose a new distributed layered negotiation based approximation algorithm to find a protecting set using local topology information. Experimental result manifests the feasibility and efficiency of our algorithms proposed.

Nonvolatile memory (e.g., Phase Change Memory) blurs the boundary between memory and storage and it could greatly facilitate the construction of in-memory durable data structures. Data structures can be processed and stored directly in NVRAM. To maintain the consistency of persistent data, logging is a widely adopted mechanism. However, logging introduces write-twice overhead. This paper introduces an optimized write-combined logging to reduce the writes to NVRAM log. By leveraging the fast-read and byte-addressable features of NVRAM, we can perform a read-and-compare operation before writes and thus issue writes in a finer-grained way. We tested our system on the benchmark suit STAMP which contains real-world applications. Experiment results show that our system can reduce the writes to NVRAM by 33%–34%, which can help extend the lifetime of NVRAM and improve performance. Averagely our system can improve performance by 7%–11%.

Dangling pointer error is pervasive in C/C++ programs and it is very hard to detect. This paper introduces an efficient detector to detect dangling pointer error in C/C++ programs. By selectively leave some memory accesses unmonitored, our method could reduce the memory monitoring overhead and thus achieves better performance over previous methods. Experiments show that our method could achieve an average speed up of 9% over previous compiler instrumentation based method and more than 50% over previous page protection based method.

Indoor localization based on WiFi signal strength fingerprinting techniques have been attracting many research efforts in past decades. Many localization algorithms have been proposed in order to achieve higher localization accuracy. In this paper, we investigate Bayes learning algorithms and some common-used machine learning algorithms. We identify a general problem of Zero Probability (ZP) which may cause significant decrease of accuracy. In order to solve this problem, we propose an Improved Naive Bayes Simple learning algorithm, namely INBS, based on our data set characteristic. INBS is applicable even though Zero Probability problem occurs. We design experiments based on off-the-shelf WiFi devices, mobile phones and well-known machine learning tool Weka. Our experiments are conducted on a floor covering 560m 2 in a campus building and a laboratory covering 78m 2 . Experiment results show that INBS outperforms traditional Naive Bayes and k-Nearest Neighbors (k-NN) algorithms and two common-used machine learning algorithms in terms of accuracy.

The network lifetime is a fundamental criterion for evaluating WSNs. Receiving and sending data packets consume energy so judicious routing strategies can effectively extend operational time. The ant colony optimization algorithm, known as ACO, is a probabilistic technique for solving computational problems. In this paper, an ACO-based routing algorithm called PACO is proposed. We take three factors into consideration. First of all, the hops are to be minimized. Additionally, energy distance is defined and needs minimizing. We also define the popularity of the sensors given the necessity to avoid those nodes that are used too often while routing. The definition of the lifetime and energy model is discussed as well. By conducting the simulation, it is proved that PACO outperforms other ACO routing algorithms, indicating that PACO is a promising routing algorithm for prolonging the lifetime of WSNs.

Atomicity violation, a common bug that obsesses multithreaded programs, has made the developing of multithreaded programs a daunting task. Atomicity is a basic correctness property of such programs. We propose strong atomicity, an efficient mechanism that helps programmers to write correct atom blocks with very little effort. We provide atom_start and atom_end for programmers to annotate an atom block. And then our runtime system will efficiently execute the block while guaranteeing atomicity. Unlike transactional memory (TM), strong atomicity permits asymmetrical atomic code regions, which means the execution of atom block not only excludes other atom blocks, but also excludes any other instructions outside atom blocks. Hence, programmers only need to simply wrap the key code region without worrying about the conflicts from outside, which saves programming effort. We have implemented our system and evaluated it with some use cases (writing or fixing multithreaded programs). With the help of our mechanism, writing an atom block is straightforward. Just wrap the block with atom_start and atom_end and it will behave like an atom operation.

Deterministic scheduling is widely used in bug detection, record & replay and deterministic execution systems which all aim to improve the reliability of multithreaded programs. This paper designs and implements a deterministic scheduling framework (DetSF) to support fast development of various deterministic scheduling systems. Based on binary instrumentation, DetSF could control the scheduling (threads' running order) of multi-threaded programs both in fine and coarse granularities. DetSF provides an efficient and easy-to-use interface for upper tools to control the scheduling of multithreaded programs. To better analyze the performance of deterministic scheduling systems, DetSF also provides a profiling subsystem to dynamically collect runtime information. DetSF could be used in quick simulation and comparing of different deterministic execution algorithms. By enforcing a certain scheduling policy, DetSF could also be applied to implement a bug detector which can manifest concurrency bugs. We demonstrate the practicality of DetSF by implementing two existing deterministic algorithms on it. With the support of our framework, it only needs to add a bit line of codes to implement the two algorithms.

Field surveillance is one of the most important applications for wireless sensor networks. Many sensors are deployed in a region of concern to detect any potential targets. On the contrary, intelligent target finds out the deployed sensors to destroy for fear of being detected and leads to defunct surveillance. Due to the importance of sensor nodes, it is necessary to hide them from the scope of target. In this paper, we introduce Proactive Stealthiness as the natural immune system covered on sensor networks. We model the stealthiness problem, and propose a distributed marking algorithm based on Breadth-First Spanning Tree. Different from previous works which only introduce self protection to wireless sensor networks, we not only maintain the connectivity of all dominating sensors but achieve the minimum hops of urgent message to gateway. This makes the method proposed more practical and sensible. Extensive simulation confirms the performance of algorithms proposed and Experimental result manifests the feasibility and efficiency.

Since the sensor node is important and critical in Wireless Sensor Networks (WSNs), it is necessary to provide wireless sensor network certain level of protection. The self protection problem focuses on using the sensor nodes of the network to protect themselves rather than other objects so that the network can resist attacks. In this paper, we study both the minimum 1-self protection and minimum k-self protection problems, and propose our distributed local optimum algorithms. In addition, we give our method of finding multiple protection sets when the network is redundant, so that the protection sets can work by turns. Experimental result manifests the efficiency and feasibility of our algorithms proposed.

Hybrid sensor network (HSN) is composed of a set of static and mobile sensor nodes. Besides the capabilities of sensing, computing and communicating like static nodes, mobile nodes have locomotion capability particularly. Area coverage is one of the most fundamental metrics for hybrid sensor network because it is crucial to the quality of monitoring. In order to improve the network coverage, we employ the mobility of mobile nodes to compensate for the uneven deployment of static sensors. In this paper we propose a 3-phase algorithm based on Hamilton Circuit. Conducted simulation shows that the algorithm can improve coverage performance with less energy conservation. Experimental results verify the feasibility and validity of the algorithm proposed.

A robust solution for context-aware multimedia delivery in the ubiquitous computing environment remains a challenging problem. Its heterogeneous and dynamic nature demands a more flexible and intelligent framework than Internet does. We propose a layered middleware solution to address this issue. The system responds to the condition of the network to offer extensibility and efficiency. Furthermore, we propose an active controlling scheme, in which the multimedia delivery is able to adapt to the environment variation in a timely fashion. In addition, we describe an application scenario of our framework, which opens out a wide prospect.

Metadata are abstraction and knowledge of wireless sensor networks and are used to provide adequate information for query processing. The purpose of metadata management is to provide adequate information for query processing, while at the same time to make the cost of maintaining the metadata as low as possible. In this paper, we discuss new issues about metadata management in wireless sensor networks; and propose a metadata management solution which includes an architecture and a model-aided approach for the base station to collect meta-data from sensor nodes. Experimental results show the effectiveness of our solution.

A wireless sensor network may consist of a large number of small, battery-powered, wireless sensor nodes and works in an unattended way. In order to manage the sensor network and collect data from the network efficiently, we need to know the state of the WSN. In this paper, we propose a model-aided approach to support the monitoring of the state of WSNs. In this approach, models are created on base station to support the monitoring of the network, and mobile agents are injected into the network to collect state information. Experimental results show the effectiveness of our approach.

A robust solution for context-aware multimedia delivery in the pervasive computing environment remains a challenging problem. Its heterogeneous and dynamic nature demands a more flexible and intelligent framework than Internet does. We propose an adaptive middleware solution to address this issue. The system responds to the condition of the network to offer extensibility and efficiency. Furthermore, we propose an active controlling scheme, in which the multimedia delivery is able to adapt to the environment variation in a timely fashion. In addition, we describe an application scenario of our framework, which opens out a wide prospect.

Intelligent city traffic for travelling navigation, traffic prediction and decision support needs to collect large-scale real-time data from numerous vehicles. As a small, economical yet reasonably efficient device, wireless sensors can conveniently serve for this purpose. In this paper, we investigate how to deploy wireless sensor networks in buses to gather traffic data for intelligent city traffic. The paper presents a self-organization mechanism and a routing protocol for the proposed sensor networks. Our work has three advantages: (1)adaptive network topology, which satisfies highly mobile city traffic environment, (2)directed data transmission, saving energy consumption of sensor nodes with limited power resource, and (3)longer lifetime because of fewer redundant network communication and balanced power usage of sensor nodes in a network.

Field surveillance is one of the most important applications for wireless sensor networks. Many sensors are deployed in a region of concern to detect any potential targets. On the contrary, intelligent target looks for the best path to traverse the sensing field for fear of being detected and leads to defunct surveillance. In this paper, we focus on how an intelligent target traverses the sensing field. We model this traversing problem, design, implement and evaluate a number of path-finding algorithms. Different from previous works which assume complete information of the sensing field, we assume that the target only can detect part of the sensor network in its detection radius. This makes the proposed methods more practical. Extensive experiments with a target and a sensor network confirm the validity of the approach.

Most sensor networks are deployed to monitor a region of interest for any potential target. An intelligent target looks for the best path to traverse the sensing field. In this paper, we model this anti-detection problem, propose a number of path-finding algorithms, and evaluate these algorithms. Different from previous works which assume complete information of the sensing field, we assume that the target only can detect part of the deployment of sensor network. This makes the proposed method more practical. Experimental results manifest the effectiveness of the path-finding algorithms.

Current deterministic systems generally incur large overhead due to the difficulty of detecting and eliminating data races. This paper presents RaceFree, a novel multi-threading runtime that adopts a relaxed deterministic model to provide a data-race-free environment for parallel programs. This model cuts off unnecessary shared-memory communication by isolating threads in separated memories, which eliminates direct data races. Meanwhile, we leverage the happen-before relation defined by applications themselves as one-way communication pipes to perform necessary thread communication. Shared-memory communication is transparently converted to message-passing style communication by our Memory Modification Propagation (MMP) mechanism, which propagates local memory modifications to other threads through the happen-before relation pipes. The overhead of RaceFree is 67.2% according to our tests on parallel benchmarks.