Side-channel attacks (SCAs) exploit power analysis to extract secret information. Researchers have employed this technique to disassemble software and retrieve cryptographic keys by examining power consumption or electromagnetic emissions. They utilized hardware or Hamming-based fluctuations measurement to profile or model the power leakage. Developers employ power modeling to comprehend software leakage, although manually profiling the power trace across various devices and architectures requires time and effort. This work proposes a custom deep learning (DL) method to model the power trace. The DL model was trained to analyze how each assembly instruction produces leakage based on its context with other instructions. The proposed method can predict the power trace with 0.9963 R² from unseen assembly instructions. This method automates device leakage testing and captures contextual and non-linear relationships to help developers understand the software behavior, significantly reducing the time and effort required for power modeling. The potential impact of this DL model on software security is that it can effectively mitigate the risk of SCAs, thus enhancing the overall security of software systems.

Bacterial pathogens contribute significantly to the global burden of disease. Understanding their complex interactions with human health is essential for developing new diagnostic, preventative, and therapeutic strategies. While recent breakthroughs have revolutionized our understanding of these relationships, the rapid expansion of microbiome research presents a significant challenge: knowledge remains scattered across scientific literature, hindering comprehensive analysis and clinical translation. To address this, we introduce MINERVA (Microbiome Network Research and Visualization Atlas), an innovative platform that leverages a fine-tuned large language model to systematically map microbe-disease associations across extensive scientific literature. MINERVA constructs a rich, ontology-driven knowledge graph that prioritizes accuracy and transparency, enabling efficient exploration and discovery of previously hidden associations relevant to clinical decision-making. The platform features specialized modules that allow researchers to analyze individual microbes and diseases, visualize complex relationships within the knowledge network, uncover hidden connections through advanced graph algorithms and machine-learning models, and perform personalized and population-level microbiome compositional analysis. These capabilities facilitate the identification of disease risks, comorbidities, and actionable insights, supporting both research and clinical decision-making. By bridging the gap between microbiome research and real-world applications, MINERVA has the potential to transform our understanding of microbe-disease interactions, accelerating discoveries and advancing patient care. The MINERVA platform is available at https://minervabio.org/.

We appreciate Dr Chirumbolo’s thoughtful evaluation and interest in our paper and welcome the opportunity to clarify MINERVA’s design choices and methodological rationale. Below, we address each point in detail. We fully acknowledge the limitation of relying exclusively on open-access sources such as PubMed and PubMed Central, as ‘explicitly stated’ in the Limitations and Future Work sections of our paper. This decision was driven by the need to ensure reproducibility, transparency, and unrestricted data accessibility. In the Limitations section, we noted: “Despite these strengths, users should remain aware of important limitations. The platform currently relies on open-access abstracts and full-text articles from PubMed and PubMed Central, which may introduce selection bias by excluding content from paywalled or non-indexed journals”. In the Future Work section, we further emphasized our plan to address this constraint: “To address these limitations and advance MINERVA, we will pursue several strategic directions. First, we aim to expand the knowledge base by incorporating additional repositories beyond PubMed, including clinical trial databases, longitudinal metagenomics datasets, and metabolomics resources. This will reduce bias and foster a more holistic understanding of microbiome compositions and their health implications” Therefore, while we recognize that a certain degree of selection bias is inherent to the current implementation, this was a decision in favor of reproducibility, accessibility, and verifiable evidence, rather than an oversight. Currently, we are actively integrating additional data sources and repositories to minimize this bias and ensure that future releases of MINERVA offer a more comprehensive and balanced representation of the microbiome–disease landscape. In this case, we also ‘explicitly acknowledge’ this limitation in both the Limitations and Future Work sections of our paper, as we are fully aware that sentence-level relation extraction does not always capture broader discourse context or co-reference dependencies. In the Limitations section, we stated: “In addition, our pipeline focuses on extracting relationships at the sentence level. This may overlook more complex associations that span multiple sentences or require co-reference resolution and discourse-level analysis, an area we plan to address in future iterations”. In the Future work section, we further noted: “We also plan to enhance the NLP pipeline by implementing cross-sentence relationship extraction and coreference resolution to capture more complex associations”. Nevertheless, we would like to note that this approach remains a widely adopted and peer-reviewed standard in biomedical natural language processing, adopted by established resources such as PubTator, SemMedDB, and several methods cited in our manuscript [1, 2]. These sentence-level models have proven to be reliable and scalable for large-scale literature mining, where interpretability and computational efficiency are critical. However, to mitigate potential false positives arising from isolated sentence interpretation, MINERVA incorporates several safeguards. Specifically, only full-confidence predictions from our fine-tuned BioMistral-AUG-7b model were included in the knowledge graph, ensuring complete internal consistency across inference runs. In addition, when multiple sentences within the same publication described a microbe–disease pair with differing relation labels, a majority-voting mechanism was used to determine the final association (Methods, Fig. 6B). Furthermore, when multiple publications referred to the same association, we explicitly calculated a relationship strength score that aggregated and weighted the evidence across studies. Finally, all extracted relations are traceable to their exact supporting sentences, providing transparency, and enabling researchers to independently assess the evidence. This combination of strict confidence filtering, redundancy validation, and evidence traceability substantially reduces the likelihood of spurious relationships. While we agree that incorporating cross-sentence and document-level reasoning would further enrich contextual understanding, we believe that the current implementation is methodologically robust, aligned with best practices in the field, and well suited for large-scale, explainable knowledge graph construction. Nonetheless, as noted in the Future Work section, we are actively developing extensions that integrate cross-sentence relation extraction and co-reference resolution in future versions of MINERVA. Again, we acknowledge the limitation of using the journal impact factor (IF) as the sole element in our weighting scheme and ‘explicitly addressed’ this in both the Limitations and Future Work sections of the paper. In the Limitations section we stated: The weighting mechanism, which prioritizes information from high-impact journals, does not currently account for other study quality indicators such as sample size, replication, or methodological rigor. In the Future work section, we further noted: Additionally, we will develop a refined weighting strategy incorporating factors like study size, methodological rigor, and replication status, moving beyond journal impact factors alone. However, it is important to clarify that the IF influences only the relationship strength score, not the inclusion or the exclusion of relationships within the knowledge graph. All associations must first satisfy strict confidence thresholds from our relation extraction model and redundancy validation across multiple sources before any weighting is applied. Consequently, while some degree of bias related to journal IF may exist, its effect on the overall graph structure and content is limited. Moreover, the inclusion of the IF was a deliberate and transparent design choice intended to modestly prioritize (logarithmic weighting) relationships consistently reported in journals with higher peer-review stringency, while maintaining a comprehensive representation of evidence from all accessible sources. As acknowledged in the Future Work section, we are currently developing a more comprehensive and robust weighting framework that integrates additional quality metrics to better reflect the reliability of the underlying studies and further minimize potential bias. We acknowledge that current performance of our prediction model (F1 ~ 71%) is limited and requires further improvement. We again, ‘explicitly acknowledge’ this limitation in the Knowledge discovery section, as follows: “The current implementation uses Node2vec embeddings, achieving an accuracy and F1-score of ∼71% on a held-out test set. The remaining ∼30% of misclassifications have important implications: false positives may lead to spurious associations, while false negatives could obscure meaningful ones. As such, predictions from this module should be viewed as hypothesis-generating rather than definitive. We encourage users to validate these findings experimentally or clinically. Future versions of MINERVA will focus on improving performance through richer node features and more advanced graph learning architectures”. As mentioned in the manuscript, the Link Prediction module is intended to assist hypothesis generation, not clinical decision-making. We understand the concern raised in this point, however, we would like to clarify that MINERVA’s risk score was designed as a literature-based heuristic, rather than as a statistical estimator. Its purpose is to reflect the direction and frequency of associations reported across studies, independent of absolute abundance values or compositional transformations. To avoid any misunderstanding, ‘we explicitly defined’ this in the Risk Assessment Module section of the paper: “In this context, ‘risk’ refers to a heuristic score that quantifies the cumulative literature-based evidence linking microbial imbalances to specific diseases. Microbial imbalances are defined as taxa whose relative abundances fall outside a configurable reference range (default: 5th–95th percentile) when compared to a healthy population. For each such imbalance, MINERVA queries its knowledge graph to determine whether the imbalance of this specific taxon is positively or negatively associated with any disease. A disease’s risk score is then calculated by summing +1 for each positively associated taxon and −1 for each negatively associated one. This simple yet interpretable scoring system reflects the direction and magnitude of literature-supported associations, offering insights into potential health concerns based on microbiome composition”. As the description in the manuscript indicates, the risk score is intended to provide a qualitative, literature-grounded indicator of potential microbiome–disease relationships rather than a quantitative measure of clinical or statistical risk. It is therefore used to support hypothesis generation and exploratory analysis, not diagnostic or inferential conclusions. We agree that formal FDR control is crucial in hypothesis-driven statistical testing. However, MINERVA’s relation extraction is deterministic once evidence thresholds are met; no inferential p-values are computed. Therefore, traditional FDR corrections are not directly applicable. In conclusion, we sincerely appreciate Dr Chirumbolo’s thoughtful engagement with our work and the opportunity to further clarify MINERVA’s methodology. We respectfully note, however, that most of the points raised do not directly apply or correspond to limitations that were already explicitly acknowledged and discussed in detail in our Limitations and Future Work sections. These aspects were transparently presented and carefully evaluated during the peer-review process prior to publication. Our goal has always been to communicate these constraints openly while outlining clear strategies for their resolution. We therefore view this exchange as a valuable opportunity to reaffirm our commitment to continuous improvement and to the ongoing development of MINERVA as a transparent, reliable, and evolving resource for the microbiome research community. The methodological limitations highlighted in the critique were already explicitly acknowledged and discussed in the original manuscript’s Limitations and Future Work sections, where we outlined their scope, implications, and planned strategies for improvement. The design choices underlying MINERVA reflect established best practices in biomedical natural language processing and knowledge graph construction, and are supported by multiple safeguards, including full-confidence filtering, redundancy validation, and sentence-level evidence traceability, to ensure accuracy, transparency, and robustness. We remain committed to advancing and strengthening the MINERVA platform, and many of the improvements suggested, such as incorporation of additional data sources, enhanced contextual NLP modeling, and refined evidence weighting, are already part of ongoing development efforts aimed at further enriching the resource. Conflict of interest: None declared. This work was supported by HEM Pharma; and AWS Health Equity Initiative, which provided credits to use their computational infrastructure to finetune the LLMs used in this study. All data, and code for reproducibility is available at https://github.com/MGH-LMIC/MINERVA. MINERVA’s platform is freely available at https://minervabio.org/. Additionally, all the entities and found relationships are available as a supplementary material in the main manuscript.

Seamless exchange of cooperative awareness messages (CAMs) and prompt wide-area propagation of emergency warnings among neighbor vehicles are essential to improve safety of vehicles in intelligent transportation systems (ITS). This paper proposes enhanced C-V2X mode-3 operations with i) efficient multiple overlapped cells with orthogonal RF resource managements, ii) smart handover across overlapped cells to minimize the signaling overhead, and iii) prompt wide-area propagation of emergency warnings across multiple cells with collaborations of eNodeBs(eNBs). In the proposed scheme, scalable overlapped cells are configured with 3 ∼ 4 orthogonal sub-channels for up to 400 vehicles per eNodeB. The eNB continuously monitors the CAM broadcasting by vehicles and assigns resources to the vehicles with balanced RF channel usages to mitigate half-duplex problem. The smart handover across overlapped cells guarantees seamless CAM deliveries even at the boundaries of cells and reduces delay by avoiding the re-registration process of vehicles at boundary of each cell. It also provides prompt propagation of wide-area emergency warnings based on collaborations of eNBs. The operations of the proposed scheme have been implemented and analyzed with NS-3 simulation, and measured results of CAM delivery rates are more than 99.4 percent in highly congested metropolitan areas. Also, the prompt propagation of emergency warnings across multiple cells are provided within 200 msec.

Cellular vehicles-to-everything (C-V2X) is an essential technology to enhance road safety, traffic management and smart mobility. C-V2X Mode 4 uses sensing based semi-persistent scheduling (SB-SPS) to mitigate packet collisions, that consists of several mechanisms (i.e., random selection of resources, probabilistic reselection, and monitoring resource utilization pattern). The packet collision probability, however, is still non-negligible, and the CAM (cooperative awareness message) delivery ratio (CDR) is far from target requirements for high reliability and low latency applications. Especially, when the number of vehicles increases beyond 300, the packet collision probability increases in the environment of congested metropolitan area. In this paper, we analyze the performance of C-V2X Mode 4 in congested metropolitan areas, and propose a balanced resource allocation scheme that is providing a remarkable enhanced performance, considering the half-duplex communication mechanism in C-V2X Mode 4. The proposed method allocates resources in distributed manner with balanced usage of subframes, instead of random selection of resources in resource map, to mitigate CAM packet losses by half-duplex communications. From the performance analysis with simulation, we confirmed that the proposed scheme provided 97% of CDR for 500 vehicles, while the existing scheme provided only 87% for the same condition.

Arterial blood pressure (ABP) monitoring may permit the early diagnosis and management of cardiovascular disease (CVD); however, existing methods for measuring ABP outside the clinic use inconvenient cuff sphygmomanometry, or do not estimate continuous ABP waveforms. This study proposes a novel deep learning model DeepCNAP for estimating continuous BP waveform from a noninvasively measured photoplethysmography (PPG) signal in real-time. DeepCNAP was designed through the combination of deep convolutional networks and self-attention. The proposed method was constructed via 10-fold cross-validation based on the MIMIC database (the number of subjects = 942, recording time = 374.43 hours). The performance of DeepCNAP was evaluated from two perspectives: estimating ABP from PPG and classifying hemodynamically unstable events (i.e., hypertension, prehypertension, hypotension, and the normal state). The mean absolute errors of the BP estimates were 3.40 ± 4.36 mmHg for systolic BP, 1.75 ± 2.25 mmHg for diastolic BP, and 3.23 ± 2.21 mmHg for the BP waveform, indicating that DeepCNAP satisfies the standards of both the British Hypertension Society (BHS) and the Association for the Advancement of Medical Instrumentation (AAMI). From the estimated BP, hypertension, prehypertension, hypotension, and the normal state were classified with 99.44, 97.58, 92.23, and 94.64% accuracy, respectively. DeepCNAP enables feasible real-time estimation of invasively measured ABP from noninvasive PPG. With its noninvasive nature, high accuracy, and clinical relevance, the proposed model could be valuable in general wards at hospitals and for wearable devices in daily life.

The number of connected vehicles are increasing rapidly in recent few years, and meeting the requirements of vehicular networks (i.e., high reliability and low latency) is challenging. The 3rdgeneration partnership project (3GPP) Mode 4 of the newly designed cellular vehicles-to-everything (C-V2X) which is specially designed for vehicle-to-vehicle (V2V) communication uses semi-persistent scheduling (SPS). SPS contains mechanisms of random selection of resources, probabilistic reselection, and monitoring resource utilization pattern to avoid packet collision. However, the collision probability especially in congested scenarios is still a challenge that makes its performance far from the expected results. In order to increase the efficiency of C-V2X, we propose cognitive collision resolution mechanism which is an additive of the standard SPS mechanism, and further we analyze the impact of different parameters on system performance. The proposed algorithm firstly senses the available resources through candidate sensing window, then chooses the candidate resource FSecondly it broadcasts the reselection counter value to reduce the collisions in congested scenarios. Result performance obtained is better than the existing mechanisms; it shows that packet reception ratio increases considerably with different number of vehicles and different values of resource reselection probability.

Intracranial hypertension (IH) following acute phase traumatic brain injury (TBI) is associated with high mortality. Objective: This study proposes a novel parameter that may identify a potentially life-threatening IH (LTH) event and designs a machine learning model to predict LTH. Continuous recordings of intracranial pressure (ICP) and arterial blood pressure (ABP) from 273 TBI patients were used as the development dataset. The pressure-time dose (PTD) and pressure reactivity index (PRx) were calculated for each IH event, and an IH event with PRx > 0 and PTD > 5 was considered an LTH event. The association between the LTH parameters accumulated over five days and mortality was analyzed. A categorical boosting (CatBoost) model was employed to predict the occurrence of a future LTH event from the onset of IH using the ABP- and ICP-related parameters. Training and validation were performed on a total of 5,938 IH events. External performance evaluation was performed in 307 IH events included in the Cerebral Haemodynamic Autoregulatory Information System (CHARIS) database. The performance of the proposed model was evaluated through the area under the receiver operating characteristic curve (AUROC). The LTH parameters were able to distinguish between the deceased and surviving patients (AUROC > 0.7, p < 0.001). The CatBoost model predicted LTH with an AUROC = 0.7 on the external test dataset. This study demonstrated that the proposed LTH prediction model has a reasonable predictive capacity for mortality. The CatBoost model anticipates whether an IH event will develop into an LTH event. The findings of this study support the usefulness of ICP monitoring.

In this paper, we present an IEEE 802.11ah (HaLow) dongle that provides sub-1 GHz wireless channel and USB (universal serial bus) interface for power supply and wired data link for easy Internet of Things (IoT) wireless networking. The HaLow dongle is implemented using CC1352P1 sub-1GHz smart RF chipset that provides 50Kbps ~ 4Mbps in 10m ~ 1Km range. The major advantages of the HaLow dongle include i) easy installation using USB interface, ii) energy efficient wireless networking with flexible channel selection and Tx rate/power adjustments, and iii) user-friendly API (application programming interface) for raw-socket-based application developments. The functional architecture of IEEE 802.11ah (HaLow) dongle is explained in detail and the performance is analyzed in practical environments.

Reliable delivery of time-critical safety messages in vehicular networks is essential in intelligent transportation system (ITS) that mainly targets to improve safety on roads. Since the vehicular networks will be inevitably overlapped in metropolitan area, efficient managements of overlapped networks with smart vehicular handovers are necessary in order to mitigate channel interferences and intensive contentions in channel accesses. Currently available standards, including IEEE 802.11p, IEEE 1609, and wireless access in vehicular environment (WAVE) standards, however, are mainly focused on the coordinated multi-channel MAC protocol operations in a single isolated ad hoc network environment without in-depth considerations on the overlapped V2X networks.In this paper, we propose a management scheme for shared control channel and network-initiated smart vehicular handovers in the overlapped V2X networks with slotted TDMA Multichannel MAC (STMC-MAC). The proposed smart handover of vehicles across WAVE basic service set (WBSS) minimized the overhead of registration message exchanges at each WBSS, mitigating packet loss in wireless communication channels. The proposed scheme had been evaluated by implementation on NS-3 network simulator, and the performances of the proposed scheme were measured and analyzed.1

Since Internet of Things (IoT) is configured to support up to 8000 various IoT devices scattered within 1 Km range, smart controls of transmission power and wireless channel according to the communication environment and channel status are essential feature for energy-efficient IoT networking. Also smart channel selection and cognitive switching are necessary in order to mitigate the channel interferences and environmental noises. In this paper, we propose a smart control for energy efficient networking of IEEE 802.11ah-based IoT with hybrid slotted CSMA/CA-TDMA (HSCT) MAC protocol. The proposed HSCT MAC protocol enhances the performance of IEEE 802.11ah on sub-1GHz RF channel with limited transmission rate. In the HSCT, the contention among IoT devices IoT devices is minimized by employing scheduled TDMA slot (T-slot), while T-slot requests are handled by CSMA/CA contention-based MAC scheme. The proposed smart RF control scheme has been implemented on TI CC1312R high-performance Sub-1GHz wireless launch pad. This paper also provides the performance analysis of the proposed scheme based on measured results.

The IEEE 802.11p, IEEE 1609, and wireless access in vehicular environment (WAVE) standards are developed to be used on multi-channels in dedicated short range communication (DSRC) spectrum. While the vehicular networks will be inevitably installed in overlapped manner in metropolitan area, most research works of VANETs, however, are focused on the coordinated multichannel MAC protocol operations in a single isolated ad hoc network environment without considering overlapped vehicular network environments that include interferences from neighbor networks. In this paper, we propose a slotted TDMA Multichannel MAC (STMC-MAC) for overlapped vehicular networks (OVN) with software-defined networking-based distribution system (SDN-DS) as infrastructure. The proposed STMC-MAC guarantees reliable delivery of basic safety message (BSM) with bounded delay and scalability on OVN with SDN-DS for metropolitan area with high vehicular density. The proposed STMC-MAC provides higher scalability, reliability, and flexibility in BSM and non-BSM data exchanges at various VN-ESS topologies and different level of vehicular density. The proposed scheme had been implemented on NS-3 network simulator, and the performances of the proposed scheme are measured, analyzed, and compared with existing approaches1.

Electroencephalography (EEG) is the primary modality for estimating the user intention in brain-computer interface (BCI). However, the suppression of the inter-subject variability (ISV) remains as a major challenge in constructing a reliable EEG-based BCI model. Subject-specific classification models have been widely used to avoid ISV, however these inherently involve time-consuming individual calibration process. This study speculated that the calibration could be minimized via clustering BCI subjects into subgroups by their respective similarity in brain power distribution at the resting state and conducted a proof-of-concept investigation. EEG recordings of twenty-nine healthy subjects from open motor imagery (MI) dataset were used in this study. K-means clustering based on brain activation in α, low β and high β-band at resting state divided the subjects into three subgroups. The efficacy of band-clustering was evaluated by comparing its MI classification performance (leftor right-hand gripping) to subject-specific and general models. Among the subjects in a cluster, ISV was lower than that in twenty-nine subjects, especially in the α-band. The MI classification accuracy using the cluster-specific model on the α-band was marked high performance (median accuracy 68.8%). The cluster-specific model had significantly high accuracy compared to general model (median accuracy = 64.6%). Furthermore, the difference of MI classification accuracy between the cluster-specific model on the α band and subject-specific model is not significant (median accuracy = 69.3%). Consequently, establishing a model by grouping clusters using similar brain activation patterns was highly beneficial for the MI classification without individual calibration process.

Aggregation-enabled wireless local area networks (WLANs) have an automatic repeat-request (ARQ) mechanism called a block acknowledgement window (BAW), which has a significant impact on frame aggregation size, throughput, and delay under noisy channel conditions. While accurate estimations of network performance are essential for traffic engineering and guaranteed quality of service, the existing mathematical modeling approaches do not include BAW operations, and thus, are not applicable to aggregation-enabled WLANs with BAW mechanism. In this paper, we propose a new and accurate Markov chain model to estimate the average aggregation size resulting from BAW operations under noisy channel conditions. The obtained average aggregation size is used in the proposed enhanced performance model of aggregation-enabled WLANs. Our numerical and simulation analysis shows that, for 50 nodes transmitting saturated uplink traffic on a channel with a packet error rate (PER) of 0.25, ignoring the BAW operations when modeling the current protocol results in throughput and access delay errors as high as 55 and 77 percent, respectively; but the proposed enhanced model, which captures the BAW's impact, produces less than 2.5 percent errors in throughput and access delay estimations.

Currently available aggregation-enabled wireless local area network (WLAN) implementations do not apply aggregate MAC protocol data unit (A-MPDU) aggregation to real-time voice over Internet protocol (VoIP) traffic because of its strict end-to-end delay requirements. When the number of VoIP sessions in WLAN increases, both mobile station (MS) and access point (AP) face heavy voice packet losses leading to unacceptable quality of service (QoS). Therefore, VoIP capacity of aggregation-enabled WLANs like IEEE 802.11n and 802.11ac is very limited under currently available implementations. In this paper, we propose a QoS-aware adaptive A-MPDU aggregation scheduler for enhanced VoIP capacity over aggregation-enabled WLANs, which adaptively applies A-MPDU aggregation to voice traffic at both MS and AP nodes, considering dynamic contention in WLAN, delay in backbone network, and end-to-end QoS requirements of real-time voice traffic. Performance evaluations of the proposed scheme showed that it can achieve as much as 5.3 times bigger capacity compared to the existing implementations, while delivering all voice packets to destination with less than 150ms end-to-end delay and less than 50ms average jitter.

This paper proposes an adaptively scaled back-off (ASB) mechanism to mitigate the performance degradations in the Binary Exponential Back-off (BEB) of IEEE 802.11 CSMA/CA in the highly dense environment such as IEEE 802.11ax high efficiency WLAN (HEW). The proposed ASB mechanism selects the optimal CW size to achieve maximized network performance adaptively based on the measured conditional collision probability (pc) and the estimated number of contending stations. The ASB protocol can provide higher efficiency than the legacy Binary Exponential Back-off (BEB) that simply adjust the back-off contention window (CW) size by blind exponential increase at repeated collision avoidance and resetting to the minimum value (CWmin) at successful transmission. The performance analysis of the proposed ASB scheme with ns-3 network simulation shows that the proposed ASB scheme can achieve 21.14% higher throughput and take 32.45% less average interval between successful transmissions than the BEB mechanism in highly dense WLANs with saturated traffic environment.1

Recently, several cluster-based time-division multiple access (TDMA)-based MAC protocols have been proposed for VANETs to ensure that all vehicles can deliver safety messages with high reliability and within a bounded delay. The existing TDMA-based MAC protocols for VANETs, however, are analyzed to be poor in scalability in terms of the number of vehicles. In this paper, we propose a TDMA-based MAC protocol that provides reliable delivery of safety messages, high throughput of non-safety messages, and high scalability where the traffic and communication densities are high in a single-hop cluster. The newly proposed scalable TDMA Cluster-based MAC (STCM) is an enhanced version of our previous Enhanced TDMA Cluster-based MAC (ETCM) that supports only up to 156 vehicles in one cluster. With simulations, we confirmed that the scalability of STCM is enhanced in various different density up to 300 vehicles in one cluster. The simulation results show that the STCM also outperforms existing MAC-layer protocols with respect to packet delivery ratio and average end-to-end delay of safety messages.

One of the challenges in QoS-aware realtime mobile multimedia service provisioning through WLANs is QoS-aware fast BSS (Basic Service Set) transition. Since most WiFi access points (APs) do not provide fast BSS transition (FT), changing currently connected AP to another takes several seconds during which mobile services are disrupted. In this paper, we propose an enhanced fast BSS transition scheme for seamless mobile broadband multimedia service provisioning and load balancing. The proposed scheme is based on i) cognitive management of WLAN ESS with SDN-based distribution system that allows checking for resource availability before selecting a candidate target AP, ii) QoS-aware roaming/handover with resource request over distribution system(DS), and iii) proactive forwarding table updates of SDN-based distribution system. The proposed scheme had been implemented and evaluated in a real IEEE 802.11n testbed environment, and the performance measurements showed enhanced QoS provisioning with minimized service disruption time and packet loss compared with existing approaches.

Currently available IEEE 802.11n implementations do not apply aggregate MAC protocol data unit (A-MPDU) aggregation to voice traffic because of its strict end-to-end delay requirements. When multiple nodes contend to send voice and other lower-priority traffic, 802.11n network faces serious performance degradation problems for two major reasons: i) due to the medium access control (MAC) and physical (PHY) layer overheads of individual short voice packet transmissions, the available network throughput for lower-priority traffic significantly decreases, and ii) increased contention results in high voice packet loss rate (PLR). This paper proposes a QoS-aware adaptive A-MPDU aggregation scheduler for voice traffic in 802.11n/ac WLANs that adaptively applies A-MPDU aggregation to voice traffic based on QoS requirements and periodically obtained average medium access delay and end-to-end delay statistics. Performance evaluations on 20 nodes sending uplink 64Kbps voice and saturated best-effort traffic at 150Mbps PHY rate showed that the proposed scheme achieved 160% throughput improvement compared to default implementations in driver. During the simulations with 50 nodes transmitting at 600Mbps PHY rate, the proposed scheme provided up to 275Mbps throughput against 0.13Mbps by default scheme; the proposed scheme delivered all voice packets (PLR 1/4 0 percent) and over 99.9 percent of them had less than 150ms end-to-end delay.

In Machine-to-Machine (M2M) communications based on IEEE 802.11ah protocol, a single access point (AP) should support up to 8000 stations within a coverage range of 1 Km using relatively low sub-1GHz wireless communication channel. The registrations of a large number of stations at network initialization or at any reboot of AP after a power failure, usually take more than 1 minute that prohibits agile cognitive sensing and adaptive control. In this paper, we propose an enhanced registration procedure for M2M communications based on IEEE 802.11ah with mitigated contention. In the proposed registration procedure, the authentication request/response and association request/response messages are extended to carry network allocation vector (NAV) to allocate the communication channel for combined processing of authentication and association to minimize contentions generated by massive wireless channel accesses from up to 8000 stations. We also analyze the optimal configuration of parameters of authentication control threshold (ACT)-based contention mitigations in registrations. The simulation experiments show that the proposed registration procedure with combined authentication and association reduces more than 15% of registration time for 8000 stations compared to currently existing solutions.

Fast fault restoration is essential in carrier-grade transport network for QoS-guaranteed service provisioning and efficient traffic engineering. The performances of multi-layered protection switching with control-forwarding separation, however, have not been studied in detail. In this paper, we propose a fast fault restoration in multi-layered network with Control-Forwarding separation. The proposed scheme leverages the merits of each layer, such as fast fault detection in WDM Optical layer, traffic engineering in MPLS layer, and hierarchical priority-based resource sharing in IP layer. The performances of the proposed fast fault restoration with control-forwarding separation are analyzed with a series of systematic tests. From the experiments, we found that the fault restoration time in IP/MPLS network with control-forwarding separation is less than 10 msec in a network with 9 forwarding elements in 3 × 3 mesh topology, while it takes around 90ms in a network with 256 nodes (16 × 16) mesh network. These results show that the multi-layered network with control-forwarding separation has scalability for carrier-grade QoS-guaranteed service provisioning.

OpenFlow switching and Network Operating System (NOX) have been proposed to support new conceptual networking trials for fine-grained control and visibility. The OpenFlow is expected to provide multi-layer networking with switching capability of Ethernet, MPLS, and IP routing. NOX provides logically centralized access to high-level network abstraction and exerts control over the network by installing flow entries in OpenFlow compatible switches. The NOX, however, is missing the necessary functions for QoS-guaranteed software defined networking (SDN) service provisioning on carrier grade provider Internet, such as QoS-aware virtual network embedding, end-to-end network QoS assessment, and collaborations among control elements in other domain network. In this paper, we propose a QoS-aware Network Operating System (QNOX) for SDN with Generalized OpenFlows. The functional modules and operations of QNOX for QoS-aware SDN service provisioning with the major components (e.g., service element (SE), control element (CE), management element (ME), and cognitive knowledge element (CKE)) are explained in detail. The current status of prototype implementation and performances are explained. The scalability of the QNOX is also analyzed to confirm that the proposed framework can be applied for carrier grade large scale provider Internet1.

This paper proposes an implementation of IEEE 802.21 MIH (Media Independent Handover) in Linux kernel space for cognitive and smart handovers. In the proposed scheme, the surrounding wireless access network information is collected and provided by IEEE 802.21 MIH, and the mobile node (MN) performs time-limited smart scanning with interleaved packet delivery for target access point. For better performance, the MIH functions have been implemented as Linux kernel modules that minimize the overheads in event notification and command processing. During the smart channel scanning,MIH is used to inform the change of MN's operation mode to network (e.g., MAG - mobile access gateway), asking the buffering of downlink packets during the smart scanning. The smart scanning is controlled to limit the service interruption time to be less than 50 ms, in order to guarantee the QoS requirement of realtime multimedia service, especially jitter and packet loss. The overall performances of the MIH operations and QoS provisioning are analyzed in detail. The proposed IEEE 802.21 MIH on Linux kernel can support various horizontal and vertical handovers across heterogeneous wireless access networks for QoS-guaranteed seamless realtime service provisioning1.

Guaranteed QoS with seamless secure mobility is strongly required in near future Internet with heterogeneous wireless access networks. In order to provide QoS-guaranteed realtime services, the packet delivery should be continued with limited packet delay, jitter and packet loss. Through the extensive performance evaluations of PFMIPv6 we found that the TCP throughput may drastically reduce due to the out-of-order packet delivery. In this paper, we propose an optimized predictive PFMIPv6 handover scheme with QoS-negotiation based on MIH (Media Independent Handover). When a target access point is selected, the timely pre-registration is provided based on MIH. We compared the performances of legacy PFMIPv6 and the proposed optimized PFMIPv6 in terms of TCP throughput. Using the optimized packet delivery between Local Mobility Anchor (LMA) and new Mobile Access Gateway (MAG) with pre-registration the proposed optimized PFMIPv6 achieved larger TCP throughput than the legacy PFMIPv6.

In this paper, we present a discussion of some architectural principles pertaining to mapping distribution protocols that are common components of proposed designs to improve the scalability of future Internet routing tables. The efficiency of a mapping distribution protocol in terms of response time and the volume of traffic load it generates are important considerations. We consider how Egress Tunnel Routers (ETRs) can perform aggregation of end-point ID (EID) address space belonging to their downstream delivery networks, in spite of migration/re-homing of some subprefixes to other ETRs. This aggregation may be useful for reducing the processing load and memory consumption associated with mapping messages, especially in resource-constrained components of the mapping distribution system. Consequently, the proposed methods can also help improve response time (e.g., first packet delay). Some key architectural issues, their potential solutions and trade-offs are discussed. The overarching goal is to expose and discuss some subtleties in design considerations for mapping distribution and management.

Correct estimation of the available bandwidth in overlapped WiFi WLANs environments is one of the essential functions for efficient network resource management and seamless mobile service provisioning of QoS-guaranteed realtime multimedia applications in future Internet. In this paper, we propose a cognitive passive estimation of the available bandwidth (cPEAB) by correct measurements of i) the proportion of waiting and back-off delay, ii) packet collision probability, iii) acknowledgement delay, and iv) channel idle time compared to measurement period. For more accurate estimation of the available bandwidth, the information of the hidden nodes and exposed nodes are provided by the cognitive network management system. Also, the proposed scheme is using passive measurements, instead of active probe packet exchange which directly affects the available bandwidth of other mobile nodes. The proposed cPEAB scheme has been implemented on Multiband Atheros Driver for WiFi (MadWiFi), and the performance has been analyzed and compared with existing schemes, such as active bandwidth measurements with probes, adaptive admission control protocol (AAC), available bandwidth estimation (ABE), and improved available bandwidth estimation (IAB). From the experimental measurements on real WiFi environment, we found that the proposed cPEAB provides the most accurate estimation of available bandwidth on the overlapped WiFi WLAN environment where the hidden/exposed nodes are dynamically affecting the available bandwidth.

In this paper we propose an IP traceback mechanism based on deterministic packet marking and logging, using protected nodes set to reduce logged data amount. The proposed scheme exploits the fact that the number of nodes that may be under attack is usually limited to a small fraction of total nodes in the Internet, greatly reducing storage requirements by logging only the traffic destined to this fraction of nodes, thus meeting the hardware limitations of high speed core routers. Before logging at the traceback-enabled router every packet is checked whether it is destined to a host in the protected nodes set by using bloom filter. Protected nodes set and list of traceback-enabled routers is managed by security management infrastructure, which can be mirrored to avoid introduction of single point of failure. Maintaining the list of traceback-enabled routers allows performing neighbor discovery in the overlay network, which is required to detect faked identification field value in IP header by an attacker. By adding initialization stage and infrastructure the proposed scheme can provide constant complexity of per-packet processing and much longer bloom filter refresh period comparing to other approaches that use logging paradigm. Performance evaluation shows that the proposed IP traceback mechanism can be implemented in the real Internet with scalability and good deployment feasibility in terms of false positive ratio and memory usage.

In this paper, we propose an enhanced cognitive resource management for QoS-aware service provisioning in wired and wireless home/office network. We enhance QoS-aware customer network management (Q-CNM) system. The Q-CNM controls overall management process in home/office network, gathers network information, and processes incoming requests through QoS-aware resource allocation with connection admission control (CAC) function. Especially the cognitive management process at the Q-CNM provides load redistribution, optimized resource utilization for QoS-guaranteed differentiated service provisioning based on obtained knowledge about network. The detailed analysis of QoS-aware resource allocation and cognitive management process at the Q-CNM are explained. The network performance and QoS parameters are analyzed based on experimental implementation of the proposed management scheme in a real testbed environment and ns-2 network simulator.

Many broadband end user terminals in home/office networks are attached to IEEE 802.11 wireless LAN (WLAN) that provides the advantages of portability. In realtime multimedia service provisioning in WLAN-based on home/office intranet, however, it is essential to have an efficient QoS-guaranteed session management scheme between QAP and STAs. In this paper, we propose an efficient home/office intranet resource management scheme using Q-SIP/SDP for realtime stream service provisioning on IEEE 802.11e-based WLAN. For the QoS-guaranteed session allocation for realtime stream services (i.e., IPTV and VoIP), the proposed intranet resource management system exchanges control messages with Q-SIP(QoS Session Initiation Protocol) proxy server that communicates with QSTAs (QoS Stations), and controls the QAP (QoS Access Point) of IEEE 802.11e WLAN. The proposed Q-SIP/SDP session management scheme is also considered for DLS (Direct-Link Setup) for making QoS-guaranteed session between STAs located in same BSS (Basic Service Set). The intranet management system is implemented using WBEM architecture, and provides resource discovery, WLAN channel allocation with CAC, and performance monitoring of each QBSS (QoS Basic Service Set).

A Grid resource management with SLA (Service Level Agreement)-based resource brokering is essential in building effective Next Generation Grid system. A SNAP (Service Negotiation and Acquisition Protocol)-based community resource broker with three-phase commit protocol has been proposed to enhance the performance of Grid resource broker, but the job execution start time is expected to take more than 80 seconds [1]. The main reasons of such long brokerage time are application level resource probing and the resource gathering without the support of tightly coupled network/system management at each domain. In this paper, we propose a Grid resource management with tightly coupled WBEM (Web-Based Enterprise Management)/CIM (Common Information Model)-based local network/system management functions. In the proposed system, the available system/network resource information is provided by the CIM repository that is updated by periodic polling and event-driven notification. The WBEM/CIM-based local management system also supports domain-specific policy-based management functions. From the performance analysis of experimental implementation, the time for Grid resource brokering can be reduced to less than 5 seconds.

Fast fault recovery depends on immediate detection of occurred fault. In some case, it is rather difficult to determine the cause of a fault in order to provide fast fault restoration/isolation. Also, service level fault, such as service level agreement (SLA) violation may not be due to physical device failure. This paper explores in detail the aspects of service level fault detection. We present the design and implementation of service level fault detection mechanism with SLA in WBEM-based inter AS traffic engineering (TE). We changed existing service location protocol (SLP) based service discovery mechanism to universal plug and play (UPnP) based one, and extended existing providers with new UPnP Provider which has not been supported by OpenPegasus WBEM. We also designed DMTF MOF (managed object format) for service discovery for SLA management in order to handle service violation. UPnP event notification mechanism, service violation handling, and C++ based provider implementation on OpenPegasus WBEM for inter-AS TE is explained in detail.

In order to provide QoS-guaranteed differentiated service across multiple domain networks, network management with inter-AS (autonomous system) traffic engineering is essential. For the inter-AS traffic engineering among different network providers, the managed objects of inter-AS traffic engineering should be standardized in public domain, and should be easily accessible by other network operators. In this paper, we propose a design and implementation of WBEMbased network management system for inter-AS traffic engineering for QoS-guaranteed DiffServ provisioning. We designed the required MOF (Managed Object Format) for inter-AS traffic engineering with inheritance from DMTF CIM MOF. The detailed design and implementation of required WBEM server/provider functions based on OpenPegasus are explained, and the performance is analyzed.

For QoS-guaranteed realtime multimedia differentiated services provisioning across multiple AS (autonomous system) domain network, session configuration with SIP/SDP and bidirectional connection establishment using UNI & NNI signaling (i.e., RSVP-TE) are essential. Also, a scalable transit networking scheme must be provided so as to configure scalable per-class-type QoS-guaranteed packet processing and to provide scalable connection admission control (CAC). In this paper, we analyze the functional architecture of session & connection management with SIP/SDP, RSVP-TE, COPS (common open policy service)-based CAC and QoS-guaranteed virtual overlay networking. We also evaluate the overall interaction procedure among functional modules for scalable QoS-guaranteed DiffServ provisioning across multiple autonomous system (AS) domain networks.

In DiffServ packet processing, three fairness problems have been emphasized in recent research works: (i) fair sharing of bandwidth among flows with different round trip times (RTT) in aggregated flows of a class-type; (ii) fairness between responsive traffic (e.g., TCP) and nonresponsive traffic (e.g., UDP); and (iii) fair distribution of excess bandwidth among aggregates. In this paper we propose a fairness-guaranteed per class-type queueing and a hierarchical packet scheduler for DiffServ-aware-MPLS networks. The proposed scheme discriminates the packet flows according to the pre-configured DiffServ class-types, and applies different packet processing schemes and a hierarchical packet scheduler to per-class-type queue. We evaluate the proposed scheme with various DiffServ-aware-MPLS networking scenarios. From the simulation results, we can verify that the proposed scheme provides a good fairness to multiple class-types under DiffServ-aware-MPLS traffic engineering.

We propose an integrated traffic engineering and management system for DiffServ-over-MPLS services in the next generation Internet, called DoumiMan (DiffServ-over-universal-MPLS Internet Manager). Using the proposed DoumiMan for DiffServ-over-MPLS network, the Internet service provider (ISP) can easily configure Diffserv-over-MPLS traffic flows among the customer's distributed sites, and can provide guaranteed end-to-end QoS by controlling the virtual topology of MPLS tunnel LSP which are configured by MPLS traffic engineering. We explain the requirements and overall operations of the DiffServ-over-MPLS traffic engineering, the architecture of the MPLS network management system, and its detailed functions for network configuration, MPLS E-LSP establishment, performance management, fault management, and MPLS-based VPN. This paper also provides the experimental results and performance analysis of the proposed traffic engineering and management system that configured and managed a sample DiffServ-over-MPLS network with Cisco MPLS routers.

We propose a packet lossless fast rerouting scheme at a link/node fault in an MPLS (multiprotocol label switching) network with a minimized buffer delay problem at the ingress node. The proposed scheme uses a predefined, alternative LSP (label switched path) in order to restore user traffic. We propose two restoration approaches: (i) an alternative LSP is initially allocated with more bandwidth than the protected working LSP, and, after failure recovery, the allocated bandwidth of the alternative LSP is readjusted to the bandwidth of the working LSP; (ii) the length of the protected working LSP is reduced by using segment-based restoration. In the proposed approaches, there is no buffer delay problem after failure recovery at the ingress node, and the required buffer size at the ingress node is smaller than the previous approach.

In order to support the mobile and multimedia services efficiently on the next generation Internet and intranet, we need a more flexible and extensible network manager. We may consider the employment of TINA (Telecommunications Information Networking Architecture), which is the software architecture to extend the capability of the IN (intelligent network), to cope with the changes of service requirement efficiently. In order to employ the TINA-based network manager for the Internet or intranet, however we need a TINA-SNMP inter-working function because the current management standard of the Internet and intranet is the SNMP. We propose an object abstract translator, which works in the TINA-SNMP gateway for the TINA-based network manager. The proposed object abstract translator is designed to increase the flexibility of the TINA-based network manager, and to assure the independency of the TINA-based network manager from the various SNMP agent implementation technologies.

In order to guarantee the user-requested quality-of-service (QoS) and keep the network utilization at maximum, it is required to manage the network performance continuously after the network installation. The performance management function should provide useful information for the network expansion and the capacity reallocation in the future. Currently, TINA provides the specification of the management functions of configuration management, connection management, and fault management; but the management functions of performance management and security management are not well-defined yet. In this paper, we propose a TINA-based performance management architecture for the efficient performance management of the heterogeneous networks or NEs with TMN and SNMP management functions. The proposed architecture is based on the distributed processing concept of TMN performance management. The proposed architecture have been designed and implemented in multiprocess and multithread structure.