Search

Fully supervised continual learning methods have shown improved attack traffic detection in a closed-world learning setting. However, obtaining fully annotated data is an arduous task in the security domain. Further, our research finds that after training a classifier on two days of network traffic, the performance decay of attack class detection over time (computed using the area under the time on precision-recall AUC of the attack class) drops from 0.985 to 0.506 on testing with three days of new test samples. In this work, we focus on label scarcity and open-world learning (OWL) settings to improve the attack class detection of the continual learning-based network intrusion detection (NID). We formulate OWL for NID as a semi-supervised continual learning-based method, dubbed SOUL, to achieve the classifier performance on par with fully supervised models while using limited annotated data. The proposed method is motivated by our empirical observation that using gradient projection memory (constructed using buffer memory samples) can significantly improve the detection performance of the attack (minority) class when trained using partially labeled data. Further, using the classifier's confidence in conjunction with buffer memory, SOUL generates high-confidence labels whenever it encounters OWL tasks closer to seen tasks, thus acting as a label generator. Interestingly, SOUL efficiently utilizes samples in the buffer memory for sample replay to avoid catastrophic forgetting, construct the projection memory, and assist in generating labels for unseen tasks. The proposed method is evaluated on four standard network intrusion detection datasets, and the performance results are closer to the fully supervised baselines using at most 20% labeled data while reducing the data annotation effort in the range of 11 to 45% for unseen data.

Beyond fifth-generation (B5G) networks aim to support high data rates, low-latency applications, and massive machine communications. Artificial Intelligence/Machine Learning (AI/ML) can help to improve B5G network performance and efficiency. However, dynamic service demands of B5G use cases cause AI/ML model performance degradation, resulting in Service Level Agreements (SLA) violations, over- or under-provisioning of resources, etc. Retraining is essential to address the performance degradation of the AI/ML models. Existing threshold and periodic retraining approaches have potential disadvantages, such as SLA violations and inefficient resource utilization for setting a threshold parameter in a dynamic environment. This paper proposes a novel approach that predicts when to retrain AI/ML models using Generative Artificial Intelligence. The proposed predictive approach is evaluated for a Quality of Service Prediction use case on the Open Radio Access Network (O-RAN) Software Community platform and compared to the predictive approach based on the classifier and a threshold approach. Also, a realtime dataset from the Colosseum testbed is considered to evaluate Network Slicing (NS) use case with the proposed predictive approach. The results show that the proposed predictive approach outperforms both the classifier-based predictive and threshold approaches.

The recent adaption of virtualization technologies in the next generation mobile network enables 5G base station to be segregated into a Radio Unit (RU), a Distributed Unit (DU), and a Central Unit (CU) to support Cloud based Radio Access Networks (C-RAN). RU and DU are connected through a fronthaul link. In contrast, CU and DU are connected through a midhaul link. Although virtualization of CU gives benefits of centralization to the operators, there are other issues to be solved such as optimization of midhaul bandwidth and computing resources at edge cloud and central cloud where the DUs and CUs are deployed, respectively. In this paper, we propose a dynamic functional split selection for the DUs in 5G C-RAN by adopting to traffic heterogeneity where the midhaul bandwidth is limited. We propose an optimization problem that maximizes the centralization of the C-RAN system by operating more number of DUs on split Option-7 by changing the channel bandwidth of the DUs. The dynamic selection of split options among each CU-DU pair gives 90% centralization over the static functional split for a given midhaul bandwidth.

To access an unlicensed channel Wi-Fi follows Listen Before Talk (LBT) mechanism whereas LTE-U adopts ON-OFF duty cycled mechanism to fairly share the channel with Wi-Fi. These contrasting mechanisms result in quite different performance for Wi-Fi and LTE-U based on their relative deployment and density in the environment. In this work, we present an analytical model for characterization of achievable throughputs of Wi-Fi and LTE-U networks in spatially distributed high-density scenarios. The proposed model is used to study how LTE-U and Wi-Fi coexist with each other in different deployment scenarios. Our extensive simulation results prove it to be a reliable model for estimating throughput of both Wi-Fi and LTE-U. We record a very good accuracy in throughput estimation and the mean normalized error is less than 1% for 40-node scenario in which 50% of nodes belong to each of Wi-Fi and LTE-U. Finally, we use the analytical model to conduct coexistence studies of LTE-U and Wi-Fi., Comment: 14 pages, 14 figures

A significant amount of research literature is dedicated to interference mitigation in Wireless Mesh Networks (WMNs), with a special emphasis on designing channel allocation (CA) schemes which alleviate the impact of interference on WMN performance. But having countless CA schemes at one's disposal makes the task of choosing a suitable CA for a given WMN extremely tedious and time consuming. In this work, we propose a new interference estimation and CA performance prediction algorithm called CALM, which is inspired by social theory. We borrow the sociological idea of a "sui generis" social reality, and apply it to WMNs with significant success. To achieve this, we devise a novel Sociological Idea Borrowing Mechanism that facilitates easy operationalization of sociological concepts in other domains. Further, we formulate a heuristic Mixed Integer Programming (MIP) model called NETCAP which makes use of link quality estimates generated by CALM to offer a reliable framework for network capacity prediction. We demonstrate the efficacy of CALM by evaluating its theoretical estimates against experimental data obtained through exhaustive simulations on ns-3 802.11g environment, for a comprehensive CA test-set of forty CA schemes. We compare CALM with three existing interference estimation metrics, and demonstrate that it is consistently more reliable. CALM boasts of accuracy of over 90% in performance testing, and in stress testing too it achieves an accuracy of 88%, while the accuracy of other metrics drops to under 75%. It reduces errors in CA performance prediction by as much as 75% when compared to other metrics. Finally, we validate the expected network capacity estimates generated by NETCAP, and show that they are quite accurate, deviating by as low as 6.4% on an average when compared to experimentally recorded results in performance testing.

The IEEE 802.11ah amendment is designed to support upto 8K M2M devices over the Sub-GHz channel. To achieve this, it introduces new modifications to the PHY and MAC layers. A dynamic Restricted Access Window~(RAW) mechanism is introduced at the MAC layer. RAW splits the access for different devices into small chunks of time. Using the RAW mechanism, we propose a novel Delay and Energy Aware RAW Formation (DEARF) scheme to support delay sensitive devices along with other delay tolerant devices. We exploit this to introduce four new RAWs. These new RAWs help split the access into contention free for delay sensitive devices and contention based for other devices. For scheduling resources between these devices, we give the DEARF resource allocation and evaluate the scheme with help of simulations. Devices suffer a higher delay but, at high loads, devices using the proposed scheme are able to save 30% energy per packet and upto 330% energy per device. Packet delivery ratio is 100% at low loads and 90% at high loads as compared to 95% and 27% given by the traditional access. Other delay tolerant devices save upto 16% on energy per packet transmitted and 23% on delay at high loads., Comment: Accepted in IEEE ANTS 2016

Achieving a tighter level of aggregation between LTE and Wi-Fi networks at the radio access network (a.k.a. LTE-Wi-Fi Aggregation or LWA) has become one of the most prominent solutions in the era of 5G to boost network capacity and improve end user's quality of experience. LWA offers flexible resource scheduling decisions for steering user traffic via LTE and Wi-Fi links. In this work, we propose a Collocated LTE/WLAN Radio Level Integration architecture at IP layer (C-LWIP), an enhancement over 3GPP non-collocated LWIP architecture. We have evaluated C-LWIP performance in various link aggregation strategies (LASs). A C-LWIP node (i.e., the node having collocated, aggregated LTE eNodeB and Wi-Fi access point functionalities) is implemented in NS-3 which introduces a traffic steering layer (i.e., Link Aggregation Layer) for efficient integration of LTE and Wi-Fi. Using extensive simulations, we verified the correctness of C-LWIP module in NS-3 and evaluated the aggregation benefits over standalone LTE and Wi-Fi networks with respect to varying number of users and traffic types. We found that split bearer performs equivalently to switched bearer for UDP flows and switched bearer outperforms split bearer in the case of TCP flows. Also, we have enumerated the potential challenges to be addressed for unleashing C-LWIP capabilities. Our findings also include WoD-Link Aggregation Strategy which is shown to improve system throughput by 50% as compared to Naive-LAS in a densely populated indoor stadium environment.

A promising solution satisfying the industry's demand to have minimum alterations in LTE for its operation in unlicensed spectrum is duty cycled LTE-U scheme, which adopts discontinuous transmission to ensure fair coexistence with 802.11 (Wi-Fi) WLANs. Even though the scheme guarantees to maintain Wi-Fi network performance, the fairness among Wi-Fi users still remains arcane. In this work, we present a practical scenario where LTE-U, despite being discontinuous (by following an ON-OFF cycle), results in not only unfair throughput distribution among Wi-Fi users but also causes degradation in Wi-Fi APs downlink performance. This is due to the domination of few Wi-Fi users who harness channel in both ON and OFF durations of LTE-U, namely non-victim users over those who get access only in OFF duration, called victim users. In this paper, we studied the performance of victim and non-victim Wi-Fi users, and Wi-Fi AP while varying LTE-U ON duration (i.e., duty cycle). A propitious scheme is proposed for WLANs, with regard to ease of implementation, employing Point Coordination Function (PCF) mode of 802.11, promising fairness among Wi-Fi users with improvement in the channel utilization of Wi-Fi network. An analytical model is developed to demonstrate guaranteed improvement and validate the simulation results., Comment: 7 pages, 9 figures, submitted to IEEE ICC 2017

The notion of Total Interference Degree (TID) is traditionally used to estimate the intensity of prevalent interference in a Multi-RadioMulti-ChannelWirelessMesh Network (MRMC WMN). Numerous Channel Assignment (CA) approaches, linkscheduling algorithms and routing schemes have been proposed for WMNs which rely entirely on the concept of TID estimates. They focus on minimizing TID to create a minimal interference scenario for the network. In our prior works [1] and [2], we have questioned the efficacy of TID estimate and then proposed two reliable interference estimation metrics viz., Channel Distribution Across Links Cost (CDALcost) and Cumulative X-Link-Set Weight (CXLSwt). In this work, we assess the ability of these interference estimation metrics to replace TID as the interferenceminimizing factor in a CA scheme implemented on a grid MRMC WMN. We carry out a comprehensive evaluation on ns-3 and then conclude from the results that the performance of the network increases by 10-15% when the CA scheme uses CXLSwt as the underlying Interference Mitigation Function (IMF) when compared with CA using TID as IMF. We also confirm that CDALcost is not a better IMF than TID and CXLSwt., Comment: 6 Pages

The recent developments in full duplex (FD) communication promise doubling the capacity of cellular networks using self interference cancellation (SIC) techniques. FD small cells with device-to-device (D2D) communication links could achieve the expected capacity of the future cellular networks (5G). In this work, we consider joint scheduling and dynamic power algorithm (DPA) for a single cell FD small cell network with D2D links (D2DLs). We formulate the optimal user selection and power control as a non-linear programming (NLP) optimization problem to get the optimal user scheduling and transmission power in a given TTI. Our numerical results show that using DPA gives better overall throughput performance than full power transmission algorithm (FPA). Also, simultaneous transmissions (combination of uplink (UL), downlink (DL), and D2D occur 80% of the time thereby increasing the spectral efficiency and network capacity., Comment: Submitted to IEEE Globecom Conference 2016

In multi-radio multi-channel (MRMC) WMNs, interference alleviation is affected through several network design techniques e.g., channel assignment (CA), link scheduling, routing etc., intelligent CA schemes being the most effective tool for interference mitigation. CA in WMNs is an NP-Hard problem, and makes optimality a desired yet elusive goal in real-time deployments which are characterized by fast transmission and switching times and minimal end-to-end latency. The trade-off between optimal performance and minimal response times is often achieved through CA schemes that employ heuristics to propose efficient solutions. WMN configuration and physical layout are also crucial factors which decide network performance, and it has been demonstrated in numerous research works that rectangular/square grid WMNs outperform random or unplanned WMN deployments in terms of network capacity, latency, and network resilience. In this work, we propose a smart heuristic approach to devise a near-optimal CA algorithm for grid WMNs (NOCAG). We demonstrate the efficacy of NOCAG by evaluating its performance against the minimal-interference CA generated through a rudimentary brute-force technique (BFCA), for the same WMN configuration. We assess its ability to mitigate interference both, theoretically (through interference estimation metrics) and experimentally (by running rigorous simulations in NS-3). We demonstrate that the performance of NOCAG is almost as good as the BFCA, at a minimal computational overhead of O(n) compared to the exponential of BFCA.

Close proximity of mobile devices can be utilized to create ad hoc and dynamic networks. These mobile Proximity Based Networks (PBNs) are Opportunistic Networks that enable devices to identify and communicate with each other without relying on any communication infrastructure. In addition, these networks are self organizing, highly dynamic and facilitate effective real-time communication. These characteristics render them very useful in a wide variety of complex scenarios such as vehicular communication, e-health, disaster networks, mobile social networks etc. In this work we employ the AllJoyn framework from Qualcomm which facilitates smooth discovery, attachment and data sharing between devices in close proximity. We develop \textit{Min-O-Mee}, a Minutes-of-Meeting app prototype in the Android platform, utilizing the AllJoyn framework. Min-O-Mee allows one of the participants to create a minutes-of-meeting document which can be shared with and edited by the other participants in the meeting. The app harnesses the spatial proximity of participants in a meeting and enables seamless data exchange between them. This characteristic allows Min-O-Mee to share not just minutes-of-meeting, but any data that needs to be exchanged among the participants, making it a versatile app. Further, we extend the basic AllJoyn framework to enable multi-hop communication among the devices in the PBN. We devise a novel routing mechanism that is suited to a proximity centric wireless network as it facilitates data routing and delivery over several hops to devices that are at the fringe of the PBN., Comment: Accepted in 2015 International Conference on Computing and Network Communications (CoCoNet'15) (IEEE)

Performance of wireless mesh networks (WMNs) in terms of network capacity, end-to-end latency, and network resilience depends upon the prevalent levels of interference. Thus, interference alleviation is a fundamental design concern in multi-radio multi-channel (MRMC) WMNs, and is achieved through a judicious channel assignment (CA) to the radios in a WMN. In our earlier works we have tried to address the problem of estimating the intensity of interference in a wireless network and predicting the performance of CA schemes based on the measure of the interference estimate. We have proposed reliable CA performance prediction approaches which have proven effective in grid WMNs. In this work, we further assess the reliability of these CA performance prediction techniques in a large MRMC WMN which comprises of randomly placed mesh routers. We perform exhaustive simulations on an ns-3 802.11n environment. We obtain conclusive results to demonstrate the efficacy of proposed schemes in random WMNs as well., Comment: Accepted in 2015 International Conference on Computing and Network Communications (CoCoNet'15) (IEEE)

The advancements in wireless mesh networks (WMN), and the surge in multi-radio multi-channel (MRMC) WMN deployments have spawned a multitude of network performance issues. These issues are intricately linked to the adverse impact of endemic interference. Thus, interference mitigation is a primary design objective in WMNs. Interference alleviation is often effected through efficient channel allocation (CA) schemes which fully utilize the potential of MRMC environment and also restrain the detrimental impact of interference. However, numerous CA schemes have been proposed in research literature and there is a lack of CA performance prediction techniques which could assist in choosing a suitable CA for a given WMN. In this work, we propose a reliable interference estimation and CA performance prediction approach. We demonstrate its efficacy by substantiating the CA performance predictions for a given WMN with experimental data obtained through rigorous simulations on an ns-3 802.11g environment., Comment: Accepted in ICACCI-2015

Wireless Mesh Network (WMN) deployments are poised to reduce the reliance on wired infrastructure especially with the advent of the multi-radio multi-channel (MRMC) WMN architecture. But the benefits that MRMC WMNs offer viz., augmented network capacity, uninterrupted connectivity and reduced latency, are depreciated by the detrimental effect of prevalent interference. Interference mitigation is thus a prime objective in WMN deployments. It is often accomplished through prudent channel allocation (CA) schemes which minimize the adverse impact of interference and enhance the network performance. However, a multitude of CA schemes have been proposed in research literature and absence of a CA performance prediction metric, which could aid in the selection of an efficient CA scheme for a given WMN, is often felt. In this work, we offer a fresh characterization of the interference endemic in wireless networks. We then propose a reliable CA performance prediction metric, which employs a statistical interference estimation approach. We carry out a rigorous quantitative assessment of the proposed metric by validating its CA performance predictions with experimental results, recorded from extensive simulations run on an ns-3 802.11g environment.

Designing high performance channel assignment schemes to harness the potential of multi-radio multi-channel deployments in wireless mesh networks (WMNs) is an active research domain. A pragmatic channel assignment approach strives to maximize network capacity by restraining the endemic interference and mitigating its adverse impact on network performance. Interference prevalent in WMNs is multi-faceted, radio co-location interference (RCI) being a crucial aspect that is seldom addressed in research endeavors. In this effort, we propose a set of intelligent channel assignment algorithms, which focus primarily on alleviating the RCI. These graph theoretic schemes are structurally inspired by the spatio-statistical characteristics of interference. We present the theoretical design foundations for each of the proposed algorithms, and demonstrate their potential to significantly enhance network capacity in comparison to some well-known existing schemes. We also demonstrate the adverse impact of radio co- location interference on the network, and the efficacy of the proposed schemes in successfully mitigating it. The experimental results to validate the proposed theoretical notions were obtained by running an exhaustive set of ns-3 simulations in IEEE 802.11g/n environments., Comment: Accepted @ ICACCI-2015

Wireless Mesh Networks (WMNs) have evolved into a wireless communication technology of immense interest. But technological advancements in WMNs have inadvertently spawned a plethora of network performance bottlenecks, caused primarily by the rise in prevalent interference. Conflict Graphs are indispensable tools used to theoretically represent and estimate the interference in wireless networks. We propose a generic algorithm to generate conflict graphs which is independent of the underlying interference model. Further, we propose the notion of radio co-location interference, which is caused and experienced by spatially co-located radios in multi-radio multi-channel (MRMC) WMNs. We experimentally validate the concept, and propose a new all-encompassing algorithm to create a radio co-location aware conflict graph. Our novel conflict graph generation algorithm is demonstrated to be significantly superior and more efficient than the conventional approach, through theoretical interference estimates and comprehensive experiments. The results of an extensive set of ns-3 simulations run on the IEEE 802.11g platform strongly indicate that the radio co-location aware conflict graphs are a marked improvement over their conventional counterparts. We also question the use of total interference degree as a reliable metric to predict the performance of a Channel Assignment scheme in a given WMN deployment., Comment: The final publication is available at Springer via http://dx.doi.org/10.1007/s11276-015-1002-4