Recently, vision foundation models have gathered wide attention. Among the many endeavors, the Segment Anything Model (SAM) makes remarkable progress toward an universal model showing unprecedented generalization ability. We propose a novel semantic segmentation model that combines SAM’s image encoder and a low rank adaptation approach (LoRA) for feature extraction and fine-tuning on aerial images. We also employ an auxiliary CNN encoder to facilitate downstream adaptation and complement for the ViT encoder on dense vision tasks. Further, cross-attention is utilized to implement feature interactions between the two encoders. Finally, the UperNet head is employed for multi-scale feature fusion and generating segmentation masks. The proposed model was evaluated on the ISPRS Vaihingen and Potsdam dataset, and achieved the best mean intersection-over-union (mIoU) of 76.44 and 78.01 for the two datasets. The evaluation results demonstrate the superiority of our model.

Logistic average misclassification percentage (lam%) is a key measure for the spam filtering performance. This paper demonstrates that a spam filter can achieve a perfect 0.00% in lam%, the minimal value in theory, by simply setting a biased threshold during the classifier modeling. At the same time, the overall classification performance reaches only a low accuracy. The result suggests that the role of lam% for spam filtering evaluation should be re-examined.

Location is an important feature for many applications, and wireless networks may serve their clients better by anticipating client mobility. As a result, many location predictors have been proposed in the literature, though few have been evaluated with empirical evidence. This paper reports on the results of the first extensive empirical evaluation of location predictors using a two-year trace of the mobility patterns of more than 6,000 users on Dartmouth's campus-wide Wi-Fi wireless network. The surprising results provide critical evidence for anyone designing or using mobility predictors. We implemented and compared the prediction accuracy of several location predictors drawn from four major families of domain-independent predictors, namely, Markov-based, compression-based, PPM, and SPM predictors. We found that low-order Markov predictors performed as well or better than the more complex and more space-consuming compression-based predictors

Location is an important feature for many applications, and wireless networks can better serve their clients by anticipating client mobility. As a result, many location predictors have been proposed in the literature, though few have been evaluated with empirical evidence. This paper reports on the results of the first extensive empirical evaluation of location predictors, using a two-year trace of the mobility patterns of over 6,000 users on Dartmouth's campus-wide Wi-Fi wireless network. We implemented and compared the prediction accuracy of several location predictors drawn from two major families of domain-independent predictors, namely Markov-based and compression-based predictors. We found that low-order Markov predictors performed as well or better than the more complex and more space-consuming compression-based predictors. Predictors of both families fail to make a prediction when the recent context has not been previously seen. To overcome this drawback, we added a simple fallback feature to each predictor and found that it significantly enhanced its accuracy in exchange for modest effort. Thus the Order-2 Markov predictor with fallback was the best predictor we studied, obtaining a median accuracy of about 72% for users with long trace lengths. We also investigated a simplification of the Markov predictors, where the prediction is based not on the most frequently seen context in the past, but the most recent, resulting in significant space and computational savings. We found that Markov predictors with this recency semantics can rival the accuracy of standard Markov predictors in some cases. Finally, we considered several seemingly obvious enhancements, such as smarter tie-breaking and aging of context information, and discovered that they had little effect on accuracy. The paper ends with a discussion and suggestions for further work.

During the recent years, research efforts have been made to study how to deploy the WLAN in a multi-hop manner. One of the many challenges to deploy a multi-hop network is how to provide delay-sensitive application support. Research have been done to study how to use DCF mode in a multi-hop network to provide delay-sensitive application support. However, few research have been made to study how to use PCF mode in a multi-hop network. In this paper, we first propose a CFP leasing scheme based on the PCF mode of 802.11 and a conservative scheduling scheme to enable voice service in a multi-hop network. Then we theoretically and numerically evaluate the performance of proposed scheme. Results show that with the proposed scheme, we can provide real-time application support for a small scale multi-hop network.

Mobile IP protocol has been proven to be problematic when it is used to support global-wide roaming due to the end-to-end signalling scheme used during the handover process which might lead to significant packet loss and generates a large amount of signaling messages. In order to solve these problems, micromobility protocols have been proposed which limit the signalling messages within a domain, i.e. a micromobility domain. The structure of the micromobility domain is directly related to the performance of the micromobility protocols. In this paper, we propose a dynamic micromobility domain construction scheme which dynamically constructs each micromobility domain according to the aggregated traffic information. Furthermore, the proposed scheme supports mobile network. Simulation results show that the proposed scheme reduces the average signalling messages significantly.

This paper studies the TCP performance over a DMT based ADSL network. The impact of DMT subchannel bit loading on the TCP throughput performance is studied. The simulation results show that there is a threshold for the signal-to-noise ratio (SNR) gap or bit error rate (BER) above which TCP throughput drops quickly. This threshold takes its value in a wide range depending on the TCP round-trip time as well as channel noise. This suggests that it would be insufficient to set a fixed target BER at, e.g., 10/sup -7/, when calculating the number of bits to be loaded in each subchannel. Instead, the bit loading should take TCP performance into account. Finally a dynamic bit loading scheme is proposed, which jointly optimizes the channel bit rate and TCP throughput performance.

A challenge for the differentiated services (DS) architecture is how to simultaneously provide end-to-end assured service (AS) for transport control protocol (TCP) sessions and the best-effort service traffic, based on a single queue management. Our research results show that both intradomain and interdomain best-effort traffic can have an adverse impact on the interdomain TCP traffic. This paper proposes a technique to achieve the desired end-to-end throughput guarantee for TCP sessions. The proposed technique is composed of a series of measures which includes: (a) a path pinning mechanism for AS allowing aggregated bandwidth reservation for AS at each intermediate router in the forwarding path, (b) a packet marking strategy, (c) a dropping policy, (d) an adaptive dropping-threshold calculation method for queue management based on aggregated reserved bandwidth and real-time traffic measurement. The simulation results demonstrate that with this technique, a high end-to-end service assurance can be achieved for the TCP traffic, while a reasonably high throughput for best-effort traffic is maintained.

This paper examines the performance of TCP, the Internet data transport protocol, over a DMT-based ADSL system. Specifically, we attempt to develop a basic understanding, using analysis and simulation, of the properties of TCP over an ADSL system. The following key results are obtained. First, high error probability leads to significant throughput deterioration. Second, there is an optimal data rate which will maximize the TCP throughput under constraint energy. We conclude the paper by presenting the preliminary analytical and simulation results.