In domains with stringent requirements for data consistency, immutability, and efficient access—such as healthcare, finance, and disciplinary inspection—existing sensitive data management systems face considerable challenges in supporting large-scale file storage and rapid multi-node access. To address these limitations, we propose a semi-converged ring network architecture combining IPFS and Geth. This architecture evaluates and compares the performance of three network topologies, emphasizing the efficiency of node discovery and data consistency in the semi-integrated ring configuration. By incorporating SQL-based querying capabilities, the proposed design ensures consistent data access across distributed locations and facilitates large file transfers. Experimental results confirm that the solution significantly enhances node discovery, data consistency, query performance, and system security.

In recent years, the rapid development of artificial intelligence and deep learning technologies has significantly advanced video analysis techniques in smart cities and their related applications, extensively used in vehicle detection, behavior recognition, and event classification. However, traditional video analysis systems typically transfer large volumes of raw data from edge devices directly to cloud servers for processing, a practice that not only incurs high communication costs but also results in significant inference delays. To address these challenges, we propose a communication-efficient edge-to-device collaborative video analysis system. This system integrates multiple edge cameras with an edge server to perform multi-camera pedestrian occupancy prediction tasks. The system efficiently extracts task-relevant features on edge devices based on the information bottleneck principle. It employs an entropy coding scheme for feature compression and transmission, significantly reducing communication costs while ensuring inference performance. Additionally, we have designed a lightweight feature extractor suitable for resource-constrained edge devices. On the edge server side, we proposed a view contribution weighting module that effectively integrates spatial information from multiple cameras and leverages historical data to enhance current inference. Through spatio-temporal information fusion, we have significantly improved the system's prediction accuracy and efficiency. Experimental results indicate that our approach achieves a better trade-off between reducing communication overhead and maintaining high performance.

Automatic medical image segmentation plays a crucial role in computer-aided diagnosis. However, fully supervised learning approaches often require extensive and labor-intensive annotation efforts. To address this challenge, weakly supervised learning methods, particularly those using extreme points as supervisory signals, have the potential to offer an effective solution. In this paper, we introduce Deep Extreme Point Tracing (DEPT) integrated with a Feature-Guided Extreme Point Masking (FGEPM) algorithm for ultrasound image segmentation. Notably, our method generates pseudo labels by identifying the lowest-cost path that connects all extreme points on the feature map-based cost matrix. Additionally, an iterative training strategy is proposed to refine pseudo labels progressively, enabling continuous network improvement. Experimental results on two public datasets demonstrate the effectiveness of our proposed method. The performance of our method approaches that of the fully supervised method and outperforms several existing weakly supervised methods.

WiFi sensing has attracted increasing attention from researchers due to its contactless and device-free characteristics. Multimodal methods based on WiFi CSI and images have made significant achievements in the field of human action recognition. However, there are also some issues that need to be addressed, such as ignoring the contribution of the subcarrier dimension in CSI and using input data with extremely asymmetric subcarrier dimension and sample size in the recognition process, and the image contains a large number of unrelated background factors about the recognition subject and is prone to leaking user privacy. To address these issues, we propose a novel multimodal action recognition method called ActionFi, which restructured the CSI into data blocks with an equal number of subcarriers and samples. Using optical flow images instead of traditional photos in multimodal learning can eliminate irrelevant background factors while preserving essential action information. We use a dual-stream network with different types of CNNs as the main body to extract the features of restructured CSI and optical flow images, and fuse these two parts of features to achieve accurate human actions recognition. We conducted extensive experiments on the large open-source dataset MM-Fi, the results show that our ActionFi achieved state-of-the-art performance in evaluation metrics.

Traffic scheduling is critical to ensuring that the switch meets required quality of service and performance goals. Existing research on traffic scheduling to improve network service quality either has few integrated circuit (ASIC) hardware implementations for switch applications or requires multiple very strict priorities. Packet scheduling has the advantage of fine control in network communications, ensuring that critical data packets or real-time data can be transmitted in a timely manner. Although the traditional fixed hardware packet scheduling method can achieve effective packet forwarding, it lacks flexibility and scalability. The emerging of the programming protocol-independent packet processor (P4) can program the scheduling algorithm into the data plane without changing the hardware, allowing users to customize packet scheduling rules according to their own needs. It can avoid resource waste, use remaining resources where users need them, and achieve a more flexible, efficient, and scalable network architecture. This paper proposes a packet scheduling algorithm implemented under the programmable data plane framework, namely Dynamic Push-In-First-Out (D-PIFO). The algorithm consists of three components: packet admission control sub-algorithm, packet sorting and priority queue adaptive mapping sub-algorithm, and intelligent queue selection strategy under different loads. The goal of this paper is to ensure that all types of traffic receive acceptable service levels while meeting strict targets for critical traffic. Simulation results demonstrate the superiority of this algorithm over existing work.

Existing work on stationary object recognition using WiFi CSI (Channel State Information) only leverages a single feature such as profile, category, etc. However, in many situations, objects have more than one feature. Multiple features can better reflect characteristics of objects and make them easier to recognize. This paper takes the first step toward multiple feature recognition using WiFi CSI. We propose WOLFE, a WiFi based object recognition framework using multiple features. Our framework matches features and labels by decoding CSI data and multi-label matrices into Gaussian latent spaces and aligning their distributions. By resampling in this Gaussian latent space, we can restore the corresponding label information from the samples and thereby making recognition. Our framework uses different pipelines during training and inference stages to achieve end-to-end recognition. In our experiments, we collect two indoor small-scale static object datasets. WOLFE achieves the recognition accuracy as high as 84.6% and 82.87% respectively. We also consider the scenarios of multiple objects and cross-domain to verify the universality of our framework. The accuracy rate in the multiple objects scenario is over 79%, and in the cross-domain experiment is around 80%, which is less than 4% decrease from the rates in their origin domains.

Meta-heuristic algorithms play an essential role in solving real-world optimization problems. However, their performance is limited by the complexity and variability of the problems. Hence, various efficient algorithms are being actively explored. The exponential distribution optimizer (EDO), having attracted attention for its efficient search performance, has been extended to several applications. However, it suffers from falling into local optima and weak exploitation. Meanwhile, it cannot be directly applied to solve binary optimization problems. To address these challenges, this paper proposes an enhanced EDO called BOMLDEDO. The Bernstein-assisted oppositional-multiple learning strategy is proposed to avoid falling into local optimality. The Bernstein-based adaptive differential strategy is developed to improve exploitation capability. Moreover, by introducing a transfer function, repair method, and binary-to-real operation, BOMLDEDO is extended to a binary version. The IEEE (Institute of Electrical and Electronics Engineers) CEC (Congress on Evolutionary Computation) test functions and engineering problems are used to evaluate BOMLDEDO's optimization performance for continuous problems. Compared to its competitors, BOMLDEDO ranks first on more than 8 out of 10 IEEE CEC 2020 functions and more than 10 out of 12 IEEE CEC 2022 functions. Meanwhile, it achieves the global optimum in 91% of engineering problems. Furthermore, the 0–1 knapsack problems are applied to verify BOMLDEDO's binary optimization capabilities, and the results show that BOMLDEDO is successfully utilized in 14 knapsack instances. The above results demonstrate that incorporating multiple strategies helps improve the performance of BOMLDEDO, making it more reliable and applicable in solving continuous optimization problems and 0–1 knapsack problems.

This article mainly investigates the model design, field programmable gate array (FPGA) implementation, and prescribed-time fault-tolerant control of a fractional-order (FO) decoupled dual-mass micro-electro-mechanical system (MEMS) gyro with deferred constraints. First, the structure of such MEMS gyro is designed to eliminate the linear acceleration in the sensing direction and its mathematical model is built based on the Lagrange’s equation. The dynamical analysis shows that such gyro can generate unpredictable, random, and disorder motions under various FOs, stiffness cross the coupling coefficients and proof masses. The designed FPGA circuit further demonstrates the undesirable chaotic oscillations of such MEMS gyro and good hardware resources utilization, avoiding the time consuming and board redesign. Second, to better solve the problems of constraints, actuator faults, uncertainties, drive couplings, and chaotic oscillations, a dependent deferred-error function superimposed to a prescribed-time function is used to guarantee no violation of constraints after a finite time. Furthermore, a$\beta $-cut type-2 fuzzy logic system (T2FLS) is employed to solve the uncertainty, and an FO hyperbolic tangent tracking differentiator (HTTD) is utilized to deal with the direct FO derivative and repeated derivative in the framework of the backstepping control. Then, a prescribed-time fault-tolerant control scheme of the FO decoupled dual-mass MEMS gyro is proposed under the actuator fault. Finally, the abundant simulation experimental results verify the feasibility and effectiveness of our scheme.

Automated medical image segmentation can assist doctors in diagnosing faster and more accurately. Deep learningbased medical image segmentation models have significantly progressed in recent years. However, the existing models fail to leverage Transformer and MLP to improve U-shaped architecture effectively and efficiently. In addition, the multiscale features of the MLP have not been fully extracted in the bottleneck of U-shaped architecture. In this paper, we propose an efficient U-shaped architecture based on Swin Transformer and multiscale MLP, namely STM-UNet. Specifically, the Swin Transformer blocks are added to skip connection of STM-UNet in the form of residual connection, which can enhance the modeling ability of global features and long-range dependency. Meanwhile, a novel PCAS-MLP module with parallel convolution is designed and placed into the bottleneck of our proposed architecture to improve the segmentation performance. The experimental results on ISIC 2016 and ISIC 2018 demonstrate the effectiveness of our proposed method. Our method also outperforms several state-of-the-art methods in terms of IoU and Dice. Our method has achieved a better tradeoff between high segmentation accuracy and low model complexity.

Wearable devices can assist users in cognitive decline through context-aware scene interpretation. They should function in real-time with sufficient functions, performance, and usability. However, the high-accuracy and low-delay scene interpretation rely on the Deep Neural Network (DNN) inference of continuous video streams, which poses enormous challenges to wearable devices because of the tight energy budget and unpredictable delay impact. In this paper, we propose a novel framework dubbed EEOKD: Energy-Efficient Online Knowledge Distillation. The framework specializes in a high-accuracy and low-cost object detection model to automatically adapt to the target video while occupying a small amount of bandwidth and tolerating changes in network delay. First, we propose efficient asynchronous distributed algorithms based on the loss gradient to alleviate the impact of delay changes. Then, we propose a novel online knowledge distillation scheme with freshness-based importance sampling and batch training to improve the generalization ability of the student model using fewer samples at lower weight updating frequencies. The new method saves energy by accelerating the model convergence and keeps good detection performance while network delay changes considerably. Finally, we implement a system prototype and evaluate its performance and energy consumption. The experimental results show that our EEOKD saves almost 15% of the energy and about 60% of the network bandwidth, and it also improves the detection accuracy by 3% on average compared with the existing approach.

Diabetic retinopathy (DR) is one of the major blindness-causing diseases currently known. Automatic grading of DR using deep learning methods not only speeds up the diagnosis of the disease but also reduces the rate of misdiagnosis. However, problems such as insufficient samples and imbalanced class distribution in small DR datasets have constrained the improvement of grading performance. In this paper, we apply the idea of multistage transfer learning into the grading task of DR. The new transfer learning technique utilizes multiple datasets with different scales to enable the model to learn more feature representation information. Meanwhile, to cope with the imbalanced problem of small DR datasets, we present a class-balanced loss function in our work and adopt a simple and easy-to-implement training method for it. The experimental results on IDRiD dataset show that our method can effectively improve the grading performance on small data, obtaining scores of 0.7961 and 0.8763 in terms of accuracy and quadratic weighted kappa, respectively. Our method also outperforms several state-of-the-art methods.

High‐performance tracking control is essential for permanent magnet synchronous motors in the perturbed environment. Given this, a new hybrid controller is proposed in this study for a permanent magnet synchronous motor with load disturbances as well as time delays. First, a new prescribed performance method is proposed to achieve the full‐state performance constraints with load disturbances. Second, a time‐varying filter is proposed for the first time to avoid the “complexity explosion” problem of the backstepping method while guaranteeing the convergence of the filtering error. Third, by combining Lyapunov–Krasovskii functionals with adaptive neural networks, the time‐delay disturbance and unknown nonlinear dynamics of the control system have been solved. The stability analysis proves that all signals in the closed‐loop system are bounded. To show the effectiveness of the intelligent controller, the comparison simulations are given to confirm the advantages of the proposed adaptive neural control scheme.

Power consumption caused by radio frequency (RF) chains in millimeter-wave (mmWave) massive multiple-input multiple-output (MIMO) systems can be solved by beam selection. However, the spectral efficiency of traditional beam selection algorithms is unsatisfactory due to the reduction in the number of RF chains and the multiuser interference. This work proposes a differential evolution (DE)-based beam selection algorithm and an improved QR precoder to reduce power consumption and increase the performance of the systems. The proposed algorithm selects the optimal beams with DE-based beam selection for each user, which reduces the power consumption and the interference among each beam. In addition, to greatly decrease the multiuser interference, we propose an improved QR precoder by equalizing diagonals and using Tomlison-Harashima (TH) theory which can greatly reduce the computational complexity and improve the performance. The simulation results show that the proposed scheme outperforms some existing algorithms in the aspects of energy efficiency and spectral efficiency.

Recent work has shown that wearable devices can assist users in cognitive decline through context-aware scene interpretation. These devices should function in real time with sufficient functions, performance, and usability. However, scene interpretation relies on the Deep Neural Network (DNN) inference of continuous video streams, which poses enormous challenges to wearable devices because the resulting computational tasks will quickly drain their batteries. Therefore, we propose an energy-efficient and content-aware DNN inference scheme to address these challenges for assistive devices. We first present the architecture of the assistive system. Then, we leverage the temporal correlation of video frames to save energy spent on the inference by designing a novel online planning method that performs Deep Reinforcement Learning (DRL) using only a subset of frames. Moreover, we come up with a lookup algorithm to select the Dynamic Voltage and Frequency Scaling (DVFS) gear scheme for further energy-saving. Finally, we implement a system prototype and evaluate its energy consumption. The experimental results show that our solution saves almost 40% of the energy on average with negligible impact on inference accuracy and latency compared with the existing approach.

Vehicle logo detection (VLD) is a special and significant topic in object detection for vehicle identification system applications. Nevertheless, the range of the research and analysis for VLD are seriously narrow in the real complex scenes, although it’s a critical role in the object detection of small sizes. In this paper, we make further analysis work toward vehicle logo recognition and detection in real-world situations. To begin with, we propose a new multi-class VLD dataset, called VLD-45 (Vehicle Logo Dataset), which contains 45000 images and 50359 objects from 45 categories respectively. Our new dataset provides several research challenges involve in small sizes object, shape deformation, low contrast and so on. Meanwhile, we use 6 existing classifiers and 6 detectors to evaluate our dataset and show the baseline performance. According to the result, our dataset has very significant research value for the task of small-scale object detection. The dataset source:https://github.com/YangShuoys/VLD-45-B-DATASET-Detection

In recent years, with the advancement of cloud computing technology and the availability of cheaper hardware, surveillance systems have become more and more common. Unfortunately, most existing systems still face many limitations, such as latency and real-time analysis issues, etc. Edge computing effectively expands the boundaries of cloud computing, migrating some computing and analysis tasks to the edge devices for execution. Edge device could perform video analysis, which may be a good solution. In this paper, we adopt the collaborative Cloud-Edge architecture to analyze surveillance video and extract video keyframes for compressing video data at the edge. Then, we provide a residual U-net neural network to perform salient object detection on the extracted keyframes. Finally, we utilize the deep reinforcement learning Asynchronous Advantage Actor-Critic (A3C) algorithm to perform the residual U-net tasks scheduling, adaptive offloading in the cloud or edge, reducing system latency, and improving real-time performance. We verified the system performance using real road surveillance videos and other public datasets. The experiment results are inspiring. It proves that the real-time processing of the surveillance video system based on a collaborative cloud-edge mechanism could obtain the optimal result within the range of tolerable latency.

Deep Neural Network (DNN) has shown remarkable success in Computer Vision and Augmented Reality. However, battery-powered devices still cannot afford to run state-of-the-art DNNs. Mobile Edge Computing (MEC) is a promising approach to run the DNNs on energy-constrained mobile devices. It uploads the DNN model partitions of the devices to the nearest edge servers on demand, and then offloads DNN computations to the servers to save the energy of the devices. Nevertheless, the existing all-at-once computation offloading faces two great challenges. The first one is how to find the most energy-efficient model partition scheme under different wireless network bandwidths in MEC. The second challenge is how to reduce the time and energy cost of the devices waiting for the servers, since uploading all DNN layers of the optimal partition often takes time. To meet these challenges, we propose the following solution. First, we build regression-based energy consumption prediction models by profiling the energy consumption of mobile devices under varied wireless network bandwidths. Then, we present an algorithm that finds the most energy-efficient DNN partition scheme based on the established prediction models and performs incremental computation offloading upon the completion of uploading each DNN partition. The experimental results show that our solution improves energy efficiency compared to the current all-at-once approach. Under the 100 Mbps bandwidth, when the model uploading takes 1/3 of the total uploading time, the proposed solution can reduce the energy consumption by around 40%.

With the development of the cellular network in the last decade, the number of IoT devices is growing exponentially and IoT applications are becoming more complex with higher requirements for Key Performance Indicators (KPIs) such as latency, accuracy and energy consumption. To address these challenges, the edge computing paradigm is often adopted to push the computing capabilities to the edge servers nearest to end-users. However, the Quality of Experience (QoE) of IoT applications is still hard to guarantee because the nearest edge servers change while users roam around. In this paper, we propose MeFILL, a Multi-edged Framework for Intelligent and Low Latency mobile IoT applications, which reduces the latencies and improves the reliability with the seamless handover of IoT devices between edge servers and leverages the Distributed Deep Learning (DDL) collaboration among edge servers. The comparison experiments show that MeFILL can effectively optimize performance KPIs of mobile IoT applications.

Visual object tracking remains a challenging task in computer vision although important progress has been made in the past decades. Particle filter (PF) is now a standard framework for solving non‐linear/non‐Gaussian problems, especially in visual object tracking. This study proposes an ant colony optimisation (ACO)‐based iterative PF for object tracking. In the proposed method, the basic idea of ACO is used to simulate the behaviour of a particle moving toward the posterior distribution. Such idea is incorporated into the particle filtering framework in order to overcome the well‐known particle impoverishment problem. An iterative unscented Kalman filter is used to design a proposal distribution for particle generation in order to generate better predicted sample states. For the likelihood model, the authors adopt the locality sensitive histogram to model the appearance of the target object, which can better handle the illumination variation during tracking. The experimental results demonstrate that the proposed tracker shows better performance than the other tracking methods.

With the rapid development of the mobile network and growing complexity of new networking applications, it is challenging to meet the diverse resource demands under the current mobile network architecture, especially for IoT applications. In this paper, we propose GANSlicing, a dynamic service-oriented software-defined mobile network slicing scheme that leverages Generative Adversarial Networks (GANs) based prediction to timely and flexibly allocate resources for IoT applications and to improve Quality of Experience (QoE) of users. Compared with the current tenant-oriented mobile network slicing scheme, GANSlicing is able to accept 16% more requests with 12% fewer resources for the same service request batch according to our evaluation. The result demonstrates that the proposed scheme not only improves the utilization of resources but also enhances the QoE of IoT applications.

The fog computing testbed can accelerate the development of fog computing. But so far, there is no real dedicated fog computing testbed to help researchers to test their prototypes, designs, and distributed algorithms in real fog computing scenarios. Researchers often verify their fog computing solutions by adapting some existing testbeds or using simulators under specific conditions, which may make their experimental results unable to withstand the examination of real production environments. This paper proposes piFogBed, the first fog computing testbed for rapid prototyping fog computing components in real environments by using Raspberry Pies and the Docker container. It can be integrated into the existing production network quickly and flexibly. To evaluate piFogBed, we demonstrate its experimental process with a medical monitoring use case, and also verify its feasibility, effectiveness and fidelity. The cost of each fog node of the proposed testbed is only 400 RMB, which is very low compared with the special fog computing equipment, so it is worth promoting and applying.

With the ever-increasing coverage of WiFi access points, WiFi offloading makes a great contribution on mitigate the gap between cellular network capacity and mobile data traffic. The cache-enabled WiFi offloading, in particular, can reduces the waiting time of mobile users by making more use of the bandwidth of WiFi links. However, it may also bring about new types of security threats. In this paper, we present a method to launch low-rate DoS attacks using the offloading architecture. We analyze the transmission process of the cache-enabled WiFi offloading and propose to insert attack traffic during the normal data transmission. The simulation results illustrate that our method can effectively produce low-rate DoS attacks, which demonstrates the feasibility of this new type of attacks and also puts forward a warning to guard against it.

Various networking applications and systems must be tested before the final deployment. Many of the tests are performed on network testbeds such as Emulab, PlanetLab, etc. These testbeds are large in scale and organize devices in relatively fixed ways. It is difficult for them to incorporate the latest personalized devices, such as smart watches, smart glasses and other emerging gadgets, so they tend to fall short in supporting personalized experiments using devices around users. Moreover, these testbeds commonly impose restrictions on users in terms of when and where to carry out experiments making them clumsy or inconvenient to use. This paper proposes to build a testbed using users' devices in their own personal area networks (PANs). We have designed and implemented a prototype of this system, which we call PANBED. Our experiments show that PANBED allows users to set up different scenes to test applications using a home router, PCs, mobile phones and other equipment. PANBED is light weighted with a size less than 16 KB and it has little impact to the other functions of the PAN. When one node keeps on sending 32-bytes packets to another for 30 seconds, PANBED exhibited little impact on the memory of the router, and the CPU load of the router was always less than 25%.

New techniques in cross-layer wireless networks are building demand for ubiquitous channel sounding, that is, the capability to measure channel impulse response (CIR) with any standard wireless network and node. Towards that goal, we present a software-defined IEEE 802.11b receiver and CIR measurement system with little additional computational complexity compared to 802.11b reception alone. The system implementation, using the universal software radio peripheral (USRP) and GNU Radio, is described and compared to previous work. We validate the CIR measurement system and present the results of a measurement campaign which measures millions of CIRs between WiFi access points and a mobile receiver in urban and suburban areas.

Perimeter distinction in a wireless network is the ability to distinguish locations belonging to different perimeters. It is complementary to existing localization techniques. A draw-back of the localization method is that when a transmitter is at the edge of an area, an algorithm with isotropic error will estimate its location in the wrong area at least half of the time. In contrast, perimeter distinction classifies the location as being in one area or the adjacent regardless of the transmitter position within the area. In this paper, we use the naturally different wireless fading conditions to accurately distinguish locations across perimeters. We examine the use of two types of wireless measurements: received signal strength (RSS) and wireless link signature (WLS), and propose multiple methods to retain good distinction rates even when the receiver faces power manipulation by malicious transmitters. Using extensive measurements of indoor and outdoor perimeters, we find that WLS outperforms RSS in various fading conditions. Even without using signal power WLS can achieve accurate perimeter distinction up to 80%. When we train our perimeter distinction method with multiple measurements within the same perimeter, we show that we are able to improve the accuracy of perimeter distinction, up to 98%.

We propose an approach where wireless devices, interested in establishing a secret key, sample the channel impulse response (CIR) space in a physical area to collect and combine uncorrelated CIR measurements to generate the secret key. We study the impact of mobility patterns in obtaining uncorrelated measurements. Using extensive measurements in both indoor and outdoor settings, we find that (i) when movement step size is larger than one foot the measured CIRs are mostly uncorrelated, and (ii) more diffusion in the mobility results in less correlation in the measured CIRs. We develop efficient mechanisms to encode CIRs and reconcile the differences in the bits extracted between the two devices. Our results show that our scheme generates very high entropy secret bits and that too at a high bit rate. The secret bits, that we generate using our approach, also pass the 8 randomness tests of the NIST test suite.

Location distinction is the ability to determine when a device has changed its position. We explore the opportunity to use sophisticated PHY-layer measurements in wireless networking systems for location distinction. We first compare two existing location distinction methods - one based on channel gains of multi-tonal probes, and another on channel impulse response. Next, we combine the benefits of these two methods to develop a new link measurement that we call the complex temporal signature. We use a 2.4 GHz link measurement data set, obtained from CRAWDAD [10], to evaluate the three location distinction methods. We find that the complex temporal signature method performs significantly better compared to the existing methods. We also perform new measurements to understand and model the temporal behavior of link signatures over time. We integrate our model in our location distinction mechanism and significantly reduce the probability of false alarms due to temporal variations of link signatures.

In a distributed heterogeneous computing system, the resources have different capabilities and tasks have different requirements. Different classes of machines used in such systems typically vary in dollar cost based on their computing efficiencies. Makespan (defined as the completion time for an entire set of tasks) is often the performance feature that is optimized. Resource allocation is often done based on estimates of the computation time of each task on each class of machines. Hence, it is important that makespan be robust against errors in computation time estimates. The dollar cost to purchase the machines for use can be a constraint such that only a subset of the machines available can be purchased. The goal of this study is to: (1) select a subset of all the machines available so that the cost constraint for the machines is satisfied, and (2) find a static mapping of tasks so that the robustness of the desired system feature, makespan, is maximized against the errors in task execution time estimates. Six heuristic techniques to this problem are presented and evaluated