Abstract In this work, we enhance the sustainability of an enterprise network (EN) by complementing it with an expert system that apprehends the explore‐exploit behavioural intelligence of Physarum to survive against the attractive‐adversarial nutritional environment. EN sustainability is dynamic since it depends on how well EN can react to an adversarial environment. We capture a reverse analogy to characterize EN's workload‐environment with Physarum's nutritive‐environment, where the high volume of workloads at the backbone network corresponds to a poor‐nutrient environment. The expert system explores EN to find out how to manage the workloads as Physarum handles its survivability, and exploits the users' workload patterns by grouping the highly communicating users together to redesign the network structure as Physarum's intelligence to exploit energy from rich‐ and poor‐nutrient food sources through redesigned tubular structures. We define two factors, such as nutrient‐intensity and chemo‐attractant to aid the redesign process. EN evolves through a set of redesigned clusters with an objective function to maximize its sustainability for a given set of explored workloads by minimizing the workloads through the backbone. EN evolution terminates when there is no change in the backbone utilization, resembling the organism's stay in a dormant state until it experiences a favourable environment. Our experimental results on an EN with a higher volume of workloads at the backbone producing 14.26 kWh energy consumption demonstrated that the developed expert system reduced the energy consumption to 11.27 kWh, thus enhanced the sustainability from 21% to 61%.

We have developed a multimodal early tonic-clonic detection (METCSD) algorithm that combines the heart rate with a few physiological vital signs for early seizure detection. Seizures, originating from abnormal brain activity, affect vital signs, such as heart rate, blood pressure, body temperature, oxygen saturation levels, and so on. Traditionally, a seizure detection is done through clinical examinations, often utilizing electroencephalograms; the detection is further enhanced using heart rate monitoring, as the heart rate escalates suddenly before the onset of seizure. However, the sudden increase in heart rate may also be due to other factors, such as anxiety, stress, and intensive physical activity, potentially leading to false positive identifications. In this paper, we have addressed the issue by considering both the deviations of heart rate over time and against norm. In addition, we have considered two more physiological vital signs, such as blood pressure and body temperature, and derived a multi-model relation to improve the early detection accuracy while lowering false positives. Our developed METCSD algorithm consists of two phases: the heart rate assessment phase followed by the physiological vital signs evaluation phase. The algorithm monitors the heart rate variability over time, identifying sudden surges indicative of possible onset of seizures, thereby enabling early detection. With gradual rise in the heart rate, it proceeds to check the selected physiological vital signs in phase 2, distinguishing between its associations with seizures and those related to stress through continuous observation. We ran the algorithm through a dataset generated from medical consultations, demonstrating an early detection of the onset of seizures with 80% (approximately) reduced false positives compared to seizure detection relying solely on heart rate variation.

Abstract We have developed an expert system comprising a self‐aware framework for resource‐efficient and accurate data transmission within a low‐power lossy sensor network (LLN) deployed for indoor monitoring. We derived both individual and group awareness, which could ensure the awareness of each sensor regarding its resources, neighbours and network environment. The proposed expert system facilitates decision‐making under dynamic environmental conditions and employs a multi‐criteria decision‐making (MCDM) model to determine the selection of the best path towards the sink node with awareness of the existing network environment. The proposed system is validated by constructing a 6LoWPAN network in the Contiki Cooja simulator. MCDM is applied to generate an adaptive objective function for the IPv6 routing protocol for the LLN (RPL) and to aid in ranking the nodes to select the best available neighbouring node, while the data accuracy is ensured by the cluster head through data correlation among its associated members. The network performance is assessed by analyzing the packet delivery rate, throughput and energy consumption against varying sensors and by comparing our proposed MCDM‐RPL with a standard RPL and a fuzzy‐based RPL, where the results show that our framework is found to be better with gains of 13, 25 and 13%, respectively.

Abstract Computing sensor systems are involved in making vital decisions based on sensed data. Thus, the data need to be trustworthy. We have made a novel attempt to analyse the trustworthiness of computing sensor systems outcomes by modelling trust as an association triangle between intelligence, transparency, and trust. Intelligence represents the computing algorithm, which makes the vital decisions, and transparency represents completeness by explicitly specifying the decision‐making process. We propose a triangulation model as a framework for assessing the trust, and we have selected 2 trust factors analogous to transparency and intelligence: the deviation of malfunctioning sensor data from its neighbours and the deviation from its own history. We have derived a set of relationships between the trust factors and their outcomes, and we have associated them with the proposed triangulation model. We also have formulated the generation of a trust subspace as an optimization problem with an objective function to maximize the trust. We have employed Tabu Search combined with Simulated Annealing to search for the best possible weighted combinations of trust factors. We have projected the experimental results onto a 3‐coordinate equilateral triangle to validate the decision analysis by displaying a definite trust or untrustworthiness of data.

We aim to present a quantification framework to assess the evolution process of an enterprise network (EN) while introducing new applications. The evolution process is defined as a series of states, where each state represents a semi-equilibrium of temporal and spatial contexts occurring during the executions of applications on network's components. The quantification of a state context is a complex task, which requires monitoring and documenting of all events occurring at all aspects of network components. Keep in mind that many of observed events may produce direct or indirect effect on the network. Thus, the evolution needs to be tracked to understand the impact of new applications on EN, such as resource provisioning and possible intrusions and attacks. We have derived a state-space model utilizing three quantifying parameters, such as node-degree, node-bandwidth utilization and node-vulnerability for now, and we utilized these parameters to quantify EN evolution microscopically by reverse engineering the output of executing applications. An experimental set up comprising of a medium-size enterprise is analyzed with two sets of web applications and the EN state change is quantified as 〈42,11.6,0.8,t1〉 and 〈76,2.7,0.3,t2〉 individually and the state change is quantified as 〈57.02,12.3,0.2,t3〉 when they executed together.

One of the core challenges in enterprise networks (EN) is how to select and manage the data storage system (DSS). In this paper, we have explored the storage architectures, such as a direct-attached storage and consolidated servers, and centralized storage architectures, such as a storage area network (SAN), network attached storage (NAS) and hybrid storage system (HSS) for EN. We have formulated the selection and management of DSS as an optimization problem, where our framework generates alterative architectures serving DSS while sustaining or enhancing EN performance. A cost model is proposed to select DSS in a cost-effective manner. The extensive simulations have been carried out utilizing the co-simulation environment, which facilitates a fine-grain outcome in performance measurement of EN and DSS using: average response time, packet loss rate, throughput, average server's utilization and service's QoS threshold. The simulation results of EN comprising of 1750 clients and 105 servers illustrate a trade-off of 77% increased in DSS's throughput utilizing HSS, 36% with SAN and 23% with NAS. However, a slight increase in response time of 2% is observed with SAN due to its complex network infrastructure in comparison with other centralized storage systems.

Abstract The bio‐organisms are resilient in nature, where resilience is viewed as a sign of intelligence for understanding and reacting to the unforeseen changes in order to retain their normal functionalities. The resilient behavior within the livings is developed through an evolutionary cycle. Here, we want to provide an enterprise network (EN) with a resilient behavior by embedding intelligent formulations to guide EN through uncertain workload, which is not considered during early design stage. We formulated the resilient management problem within EN as an optimization problem, with an objective function is to maximize the resilience. We developed a bio‐inspired tool comprising of Genetic Algorithm (GA), Molecular Assembly (MA) and Ant Colony Optimization (ACO). Our experimental results on an EN of 100 nodes, with a heavy backbone traffic of 7.6 TB demonstrated that GA gains the maximum resilience of 50%, whereas MA gains 46%, ACO gains 31.54%, and the original EN had 20% resilience. However, with light workload, MA and GA performed closely with a maximum resilience of 64% and 62%, respectively, and ACO showed a moderate resilience with a maximum of 32%.

The security and trustworthiness of data transmitted by sensors within the wireless sensor networks (WSN) are highly crucial, since the sensors' data can be easily accessed and manipulated by intruders or malfunctions. Data security within WSN can be viewed as how the data is protected during transmission, whereas the data trustworthiness represents the confidence of the sensed data. The unfriendly operational environment, insufficient energy, and hardware malfunction may be the cause of sensors' erratic behaviors, thereby making the sensed and collected data unreliable. Thus, data trustworthiness should ensure the confidence of the sensed data. In this work, we have developed an analytical model to represent the data trustworthiness of WSN, which is a first-order polynomial combining two trust factors. The trust factors are derived based on the average value of the previously transmitted data from the sensor, and the average value of the similar sensing data reported by its neighboring sensors. We have developed an algorithm to find the sensible subspace to operate on ensuring the maximum trust, where Simulated Annealing is utilized as a search tool to select the best weighted combination of trust factors. The experimental outcomes with varying deviation of sensing parameters reveal that 67% of the estimated trust-scores are in acceptable region.

We present a novel green communication framework incorporating the redesign of existing campus enterprise network (CEN) to offset the carbon emissions at the backbone based on the theory of data encoding and power spectral density (PSD) estimations. CEN is viewed as a collection of service nodes distributed in various buildings, wherein the nodes are flexible to be clustered for serving a specific application. The carbon-offset at the backbone is indispensable as it is often susceptible to heavy traffic flow, thereby necessitating high-power cooling equipment to reduce the intensive heat spots generated from the equipment, such as routers and servers. The proposed framework accounts carbon emission from the average power consumed by the transmitted data through integrating the area under PSD curve of the encoded-transmitted data. The redesign problem is formulated as an optimization problem to redistribute the heavily communicating nodes at the backbone and to dispense the heat spots away from the backbone. We have utilized two bio-inspired algorithms such as Genetic Algorithm (GA) and Simulated Annealing (SA) to search the redesign space. The simulation results with the Manchester encoder within GA offer a maximum reduction of 23.6% of annual carbon emissions when compared with the initial CEN, whereas the SA reduces the carbon emission by 16%.

Lately, the business enterprise networks (BEN) are expected to support more online applications than what are initially designed for, due to the competitive nature of the business nowadays. The business applications are known to generate sudden increase in traffic workload, which poses challenges in providing guaranteed service to all the new and old clients. Thus, the network administrators are faced with many challenges in how to smartly manage BEN with the available resources and at the same time maintain satisfactory returns on the business. Over-provisioning of resources may not be a feasible solution, and thus, BEN can be improved by re-synthesizing the existing infrastructure by localizing the workflow. At the duration of the sudden increase in the workflow within BEN, we view the infrastructure as a fully connected graph and the aim is to convert the fully connected graph into a connected one by re-grouping the clients into new clusters. We have formulated the conversion problem as an optimization problem with an objective function to maximize the local traffic within the new clusters; moreover, we have utilized Ant Colony Optimization (ACO) to find the suitable conversion. The simulation results show that re-synthesizing a typical BEN comprising of 100 clients with a heavy traffic of 7.54 TB into a set of 5 clusters reduces the backbone traffic by 18%.

The proposed soft computing tool is developed as a suite of software programs capable of envisaging the evolution within wireless sensor networks (WSN) by accomplishing a series of processes. These processes are: modelling a real problem scenario in computing form, formulating the problem as an optimization problem, evaluating the outcomes and finally, presenting the outcomes in a statistical form. We have examined many possible failing scenarios of WSN operations through the proposed tool, which was capable to restore WSN in the presence of failures to continue normal operations. Simulated Annealing (SA) is utilized as a search method, which is embedded within the soft computing tool to explore the restoration space comprising of possible alternative topologies generated for each failing scenario, taking lifespan into consideration. The tested scenarios and their restorations can be hard-coded into a real WSN, thus preparing WSN to evolve through failures. The simulation results of a typical WSN illustrates the capability of the proposed computing tool to restore 79% of the lost days when no restoration scheme was applied.

Web storage systems are capable of providing a single storage area for managing multiple types of unstructured information in an enterprise infrastructure. Recently, network storage architectures such as network attached storage (NAS) are emerged as shared, adaptable and high-performance web storage systems for data intensive distributed applications. NAS provides filelevel data storage services to the attached heterogeneous group of clients. Efficient implementations of NAS can have a significant impact on the overall network performance and its workload. In this paper, we have proposed a simulation model integrating NS-2 and Matlab/Simulink to evaluate the performance of NAS workload and its underlying network. NS-2 works as a master simulator, and it is utilized to model and simulate the data transfer over the distributed network. Matlab/Simulink is utilized to model and simulate the data process within the web storage systems. The performance of web storage system is evaluated through the average response time of the clients' requests of the data network before and after implementing NAS. For a small-size network of 100 clients and 4 servers, our simulation results show that NAS implementation improves the average response time of the clients' requests by 31% in comparison to the existing direct attached disks (DAS).

The deployed base stations within a wireless sensor network (WSN) serve as an intermediate node in connecting the sensors to the central server. The base stations are vulnerable to malfunction failures, which may disconnect a set of sensors linked to the failed base station. This disconnection decreases the connectivity and the lifespan. In this paper, we have considered the WSN operations during base station failures and we have proposed a communication restoration framework to restore the failed communications without adding new base stations. We have considered the layout of WSN as a linear grid platform, whereby the sensors are deployed at the center of each cell and the base stations are deployed at the cell intersections. The sensors are connected in multi-hop tree topology and are grouped into various clusters with each cluster head connected to a nearby base station. On failure of any of the base stations, the restoration framework distributes the disconnected sensors to a nearby base station and if necessary, it relocates the base station to the proximity of the failed one to improve the lifespan and connectivity of WSN. Simulated Annealing aids in relocating the base stations to better locations with minimal energy consumption. The experimental results show that for the given medium size WSN with 25 sensors and 5 base stations with base stations failing consecutively at regular time intervals, the restoration model manages to maintain 100% network connectivity at the expense of 25% reduced network lifespan.

We have exploited some of the features of molecular assembly (MA) process combined with artificial intelligence (AI) to manage three network design problems occurring simultaneously, namely nodes clustering, network hierarchy and firewall placement. These exploitations have been transformed into a design tool to manage the attraction forces within all the node-pair of an enterprise network to amalgamate highly associated nodes into clusters, thereby alter the given single clustered network into a secured hierarchical network. The proposed design tool is comprised of two phases; discovery and assembly, where the discovery phase explores and analyzes the forces of attractions between nodes; moreover, the assembly phase ensures that the network is completely assembled and optimized. The simulation results illustrate the capability of the tool to assemble the 50-node enterprise network into 3-layered network in few minutes. Therefore, the tool manages to reduce the external traffic at the backbone by more than 40% in comparison with a non-molecular assembly algorithm, such as Genetic Algorithm (GA).

SUMMARY We have viewed an enterprise information network as a flexible topology, where each node seeks the opportunity to join a cluster having high association with its nodes to reduce the external bandwidth and thereby producing less carbon. The opportunistic clustering is modeled as an optimization problem and we utilized heuristics such as genetic algorithm (GA) and simulated annealing (SA) to search for the best opportunistic topology. We have proposed redesign operations such as move and swap nodes to change the existing topology into an opportunistic topology. Also, we defined an opportunistic factor ranging from +1 to –1 to measure the gain in bandwidth and the amount of reduction in carbon emissions. Our simulation results demonstrate positive opportunistic factor in individual clusters from 0.4 to 0.7 in move operation, thereby showing a total bandwidth gain of 21% and 19% within GA and SA, respectively. In the swap operation, the opportunistic factor reached a maximum of 0.3, thereby showing a bandwidth gain of 9.7% and 5.1% within GA and SA, respectively. It is also observed that the optimization within GA outperforms SA in offsetting carbon emission with a maximum of 22%. Copyright © 2012 John Wiley & Sons, Ltd.

We have proposed a design tool for synthesizing multitier computer networks based on the concept of molecular assembly (MA), where the network's nodes integrate intelligently together by exploiting various forces of attraction existing between the nodes. Three forces are defined and the forces are the distance force, the incoming flow force and the outgoing flow force. Our simulation results demonstrate that for a given unassembled network of 50 nodes, our design tool forms a self-assembled network and manages to reduce the traffic at the backbone by 40% in a short computing time.

Most of the current research in wireless sensor network (WSN) focuses on energy related issues because of the fact that tiny sensor devices possess a limited power supply. Out of the three normally executed sensor's tasks (sense, process and transmit), data transmission consumes most of the power. In this paper, we propose a query-based data aggregation model that is based on the base stations within WSN that query the sensors to transmit their collected data due to special events. The worst-case scenario for a query-based activation would be that all sensors transmit their collected data simultaneously to the base station. This could lead to a loss of data due to the overlapping of transmissions at the base station. Therefore, we have embedded our query-based model within a Monte Carlo Simulator to explore the best- and worst-case scenarios for a base station to initiate its queries to all sensors. Monte Carlo Simulator is utilized to evaluate the throughput, which is the amount of data collected at the base station, under three schemes; contiguous aggregation, aggregation with overlapping of sensing tasks, and aggregation with overlapping of sensing and processing tasks. Our simulation results demonstrate that, for the WSN of 25 sensors with a single base station deployed within the WSN, the aggregation scheme with overlapping of sensing and processing tasks shows better performance by aggregating a minimum of 56% of the data in lower time duration in comparison to other schemes.

In this paper, we have proposed an energy constrained algorithm which selects direct transmission or multi-hop transmissions based on the residual energy level of the transmitting sensor node. The sensors with higher energy levels can directly transmit their collected data to the base stations, while the sensors with low energy level can employ a two-hop transmission to reach the base stations. Our proposed data management algorithm rules out the selection of hotspot sensors, which are located closer to the base stations, as the intermediate sensors to avoid the dying (battery run out) of these nodes. The proposed data management algorithm selects one of the neighborhood nodes having minimal Euclidean distance and maximum energy level as the intermediate node for transmission. We have also added constraints to avoid the repeatability in the selection of neighborhood sensors. We have utilized as soon as possible (ASAP) and as late as possible (ALAP) scheduling algorithms in combination with the proposed data management algorithm to manage the data collection from the sensors to the base station. The simulation results show that the proposed data management algorithm performs better in reducing the energy consumption and waiting time of the selected sensors compared to the direct transmission.

This paper proposes a novel data-aggregation approach for capacity planning of a wireless sensor network (WSN). The approach is based on incorporating the three sensor's tasks, which are sensing, processing and transmission, into a task flow graph (TFG); moreover all TFGs within WSN are merged into one super task flow graph (STFG). Also, we have modeled the execution time of transmission task as non-preemptive imprecise computation times. In real time systems, the insufficient number of gateways may partially terminate the transmission due to the deadline requirements in timings. This termination of transmission of any sensor node results in partial information of the sensed data resulting in imprecise computations. We have scheduled all tasks within STFG to determine the number and capacity of the gateways (base stations), where the aggregated data should be collected. We have utilized a Branch-and-Bound algorithm to perform scheduling, which is subject to concurrency in transmissions. We have analyzed the performance of 50 sensors within WSN by varying the availability of gateways. The computational results have provided excellent bounds on the number and capacity of gateways keeping in mind the trade-off in the quality of data-aggregation in the imprecise computations.

In this paper, we have described a redesign methodology for an existing network in the educational system of 6--9 levels in the State of Kuwait, through servers consolidation. We have examined a typical school network, where the installed five servers are underutilized by more than 44% of the time. The redesign methodology operates offline by taking a snapshot of the existing network with its clients, servers and their interconnections and workloads, and translating them into a model to be manipulated for better performance with the reduction of number of servers. Our technique searches for the best possible ways to consolidate the existing servers, while maintaining an acceptable performance and increasing the utilization of remaining servers. The initial experimental results show improvement in the utilization from 44.2% to 73.61% for a network originally comprised of five servers serving 40 clients and consolidated to three servers.