Novel Baseline Computational Predictive Model for Seamless Transmission in 5G

doi:10.21203/rs.3.rs-2495762/v1

Download PDF

Research Article

Novel Baseline Computational Predictive Model for Seamless Transmission in 5G

https://doi.org/10.21203/rs.3.rs-2495762/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

5G technologies is known for its beneficial characteristic of supporting largely connected network and high-speed data transmission. With increasing number of services and application meant for hosting over 5G network, there is also an increasing concern towards accomplishing better quality of service and quality of experience in global telecommunication sector. In this line of communication advancement, it is also noted that machine learning is one of the contributories and enabling technology towards boosting the performance of value-added services and applications running on 5G networks. Existing review of literature exhibited multiple variants of methodologies meant for performing predictive performance towards leveraging quality of data delivery services in 5G. However, there are quite many research challenges too that is directly associated with deploying a cost-effective learning scheme in 5G. Therefore, the proposed scheme contributes towards developing a novel and yet simplified baseline architecture which targets to accomplishing seamless and reliable data dissemination services in 5G. The proposed model constructs a novel deployment scenario where a user handheld device is considered as a mobile node with an agenda to considered routing in allocated multiple paths to reach its destination. Further, RFC 8822 is used for deployment 5G standard along with a specific mobility model sync with a real-time server via access point and gateway node in large deployment scenario. Finally, an enhanced Long Short-Term Memory is implemented towards performing identification of predictive routes that are shared to users directly over 5G network. The study outcome is benchmarked with existing learning schemes to exhibit that proposed scheme offers approximately 35% of reduced losses, 19% of higher throughput, 23% of reduced delay, 37% of reduced memory, and 41% of reduced processing time.

Machine Learning

Quality of Service

Quality of Experience

Long Short-Term Memory

Mobility Model

Learning

5G technology is essentially meant for offering superior user experience with higher consistency, maximized availability, enhanced capacity to support massive network, highly reliable, supportability towards ultra-low latency, along with higher rate of data transmission speed [1][2]. However, the importance of the 5G technology is not due to the above-mentioned characteristic but on its power to transform the world of communication system [3]. One of the direct influences of 5G is its potential support towards Internet-of-Things (IoT) and Internet-of-Vehicle (IoV) [4][5]. Usage of IoT enables the availability of a massive set of interconnected networks that not only assists in perspective of high-end communication but also assists in opening avenues of various analytical schemes to deal with big data [6][7]. Further IoV targets not only to make the transportation system intelligent but also to ensure higher ranking of user experience while it further pave the way to futuristic unmanned autonomous vehicle [8][9]. Apart from this, 5G also potentially supports the industry to leverage the principle of automation adhering to maximize the adoption of Industry 4.0 [10][11]. Owing to its superior data transmission rate and sustainable bandwidth, 5G network can assists in establishing far reaching connectivity which conventional connectivity cannot reach along with its significant technological revolution in augmented reality [12], virtual reality [13], 3D holograms [14], etc. 5G also improves upon the sector towards security [15] and health [16]. However, irrespective of above-mentioned beneficial features, 5G is also shrouded with various challenges [17]. The primary challenge in 5G network is associated with its hosting of multiple services with massive number of operating devices, technologies, and heterogeneous networks [18]. The biggest challenge resides in introducing standardization of large services to cater up dynamic requirement of users. Apart from this legislation of cyberlaw, rising concern of privacy and security, integrated upcoming application with sensing, navigation, communication, and massively connected infrastructure is yet a wide-open challenge in 5G technology [19][20]. All this above-stated challenges are required to be solved in order to for the 5G to be completely harness its potential towards various application.

Since the past half decade, there is an exploration about the potential of Artificial Intelligence and Machine Learning approach in leveraging 5G technology [21]. The significance of machine learning is towards its capability to use massive data subjected for computation analysis to yielding efficient decision based on input data [22]. When machine learning is used in networking than it assists the network engineer to generate intellectual analytics with better predictive capability towards understanding the flow of traffic which directly can control many performance metrics without any intervention of human [23]. From the perspective of 5G network, machine learning can potentially assist in performing proactive and predictive operation that maximizes the value of using 5G networks and its supported services and applications [24]. Apart from this, it also assists in automating various business process with enhancement towards business enterprises thereby understanding the need of customers. It is to be noted that 5G offers an exponential connectivity to a large number of heterogeneous devices that runs over underlying complex forms of network and computing architectures [25]. This eventually constructs a high-end capability towards analyzing big and complex form of data in order to curve it for better value-added services or products. However, the prime focus on integrating machine learning approach is to optimize the network performance along with minimization of the capital expenditures. However, this is not so simpler task to implement such integration.

On the contrary, the data generated by the machine learning has witnessed an increased utility towards curving a predictive tool which can intrude the privacy information of the user [26]. Apart from the perspective of security challenges, the bias can be quite amplified by machine learning giving rise to higher iterative and complex process to check and minimize it. Apart from this, adoption of either supervised, unsupervised, or even hybrid form of machine learning approaches is always accompanied by both beneficial and limiting characteristic [27]. At the same time, deep learning has penetrated more in improving predictability performance in 5G. However, there is a potential difference between adoption of either of this technique. Deep learning approach does not have any dependency towards network data structure while machine learning does. Apart from this, owing to inclusion of massive number of protocols and services in 5G, there is possible occurrence of errors. Such errors can be autonomously checked by deep learning without any intervention of humans which is not the case with machine learning. However, a deeper insight towards trends of Artificial intelligence shows that machine learning is more considered in investigating 5G network as it can function using low resources which is not the case with deep learning. Apart from this, execution time of machine learning is significantly lower compared to deep learning. Therefore, the proposed scheme implements a novel scheme of machine learning approach which can optimize the performance of traffic management in 5G. The contribution of proposed scheme are as follows:

Proposed scheme deploys a large and novel scenario of 5G service deployment meant for facilitating navigational services to mobile user via their user handheld device.
An enhanced Long Short-Term Memory is used for constructing a predictive model towards furnishing distributed information of optimized routes.
The study outcome is benchmarked with existing learning scheme on various parameters to show its effectiveness.

The organization of the paper is as follows: Section 2 discusses about the recent methodologies towards machine learning in 5G followed by highlights of research problems in Section 3. Methodology is discussed in Section 4, system design in Section 5, result is discussed in Section 6 and Inference of outcome is carried out in Section 7. Finally, conclusive remarks are stated in Section 8.

This section outlines various recent studies where machine learning has been used to address various sorts of problems associated with communication deploying 5G. The work carried out by Bega et al. [28] have developed an analytical model towards admission control with network slicing using a Markov model that can obtain the best acceptance policy in 5G network. Adoption of machine learning is also witnessed in 5G based vehicular network using clustering algorithm as noted in work of Koshimizu et al. [29]. The technique has used Support Vector Machine for obtaining the accuracy in clustering process specifically aiming towards autonomous driving system. Further study towards improving vehicular network was reported in work of Tayyaba et al. [30] where a framework for policy is constructed using flow-based methodology where the communication is allocated with multiple radio resources. Strodthoff et al. [31] have used machine learning towards improving Hybrid Automatic Repeat ReQuest (HARQ). The prime agenda of this model is to carry out prediction for performing decoding process associated with the wireless 5G communication with reduced latency. The work carried out by Gheisarnejad et al. [32] have used fuzzy logic integrated with deep reinforcement learning for stabilizing the voltage requirement for power systems in 5G. The work of Polese et al. [33] have used machine learning towards design development of a unique controlling system for edge network in 5G. The idea of this work is to predict optimal number of users within a specific base station targeted towards navigational support in vehicular network. The work of Kucuk and Uysal [34] have carried out an investigation towards identifying anomaly in 5G network using self-organizing network. Han et al. [35] have carried out an investigation towards determining the beam index in 5G cells. A unique assessment model is developed using Software Defined Radio where a channel learning technique has been used. Ramachandran et al.[36] have addressed the challenges associated with the network management of 5G where a unique analytical model has been constructed using machine learning. The contribution of this model is towards facilitating 5G-based automation of closed loop for accomplishing resource utilization. Zolotukhin et al. [37] have carried out a study towards vulnerabilities associated with different machine learning approach. Yu et al. [38] have used federated learning integrated with reinforcement learning targeting towards improving the resource management in 5G used with edge computing. The model contributes towards controlling the network resources as well as delay due to offloading in order to improve the task execution performance in 5G.

Existing system has also witnessed the adoption of machine learning towards 5G network as seen in study of Saha et al. [39]. The study contributes towards using machine learning for analyzing attack graphs over a virtualized environment of software defined network in 5G. Nguyen et al. [40] have used reinforcement learning towards managing user plane function for formulating optimized network stochastic policy. The work carried out by Hernandez et al. [41] have used machine learning approach considering optical network, which is another form of network supportive of 5G system. According to this model, a supervised machine learning model is used for learning the delay network to offer accurate data. Sevgican et al. [42] have developed an intelligent model towards performing a smart data analytical operation for 5G data. The study has used multiple learning schemes for predicting capability of network load. The work carried out by Morgabo et al. [43] have addressed the challenges associated with resource management of radio using spectrum sharing. The study has used reinforcement learning towards performing predictive analysis of network slices of 5G network. Coutinho and Boukerche [44] have used transfer learning towards overall management and control of Industrial application on large scale in 5G. Adoption of reinforcement learning is also reported in work of Troia et al. [45] where embedding of virtual network as well as admission control is carried out. The study also uses integer linear programming as well as heuristics in order to enhance the performance of resource allocation. Tashan et al. [46] have presented a discussion of frequently used machine learning approaches towards optimizing functions of mobility-based applications in 5G. The study outcome contributes towards a finding that suggest that still there are open end problems associated with perspective of inclusion of mobility. The work of Rony et al. [47] have used machine learning towards allocation of dynamic spectrum for assisting in smart traffic management in 5G networks. Therefore, there are various schemes of machine learning that has been proven fruitful towards improving the performance of 5G by addressing various challenging issues associated with its applications and running services. The next section discusses about research problem.

From the prior section, it is noted that there are various approaches of machine learning towards improving the data transmission performance in 5G network. Majority of the techniques has considered a specific use case of application to be solved using different variants of machine learning algorithm towards accomplishing claimed benefits. However, there are open-end research problems in existing system briefed as follows:

Existing schemes doesn’t facilitate cost-effective computational model of machine learning approaches towards large scale and heterogeneous nodes in 5G.
None of the existing machine learning models has reported of need of change in channel in order to adjust the dynamic demands of 5G traffic.
Dependencies of iterative operations as well as resource consumption during training operations is not much reported to be addressed in existing schemes.
Assessment of computational complexity and its possible influence in data transmission performance is also not addressed for large and complex network in 5G based application/services.

Therefore, the proposed research work addresses the above-mentioned challenges in machine learning which are found to be manifold and interconnected. The next section outlines the solution presented by the proposed scheme.

The prime goal of the proposed scheme is to develop a novel and simplified computational model which can predict the density of the users and thereby facilitate them towards a proper navigational in traffic system. The applicability of the proposed study is towards any form of 5G based network which includes dynamic mobility. The accuracy in defining the navigational path is contributed using enhanced Long Short-Term Memory (LSTM). Further information of the proposed implementation is as follows:

4.1 Adopted Methodology

The first and foremost step of implementation is to deploy RFC 8822 [48] for the purpose of deploying 5G architecture followed by developing a mobility model and predictive model using deep learning approach. From the prior section, it has been noted that there are various artificial intelligence-based scheme deployed in 5G for addressing multiple problems. The prime agenda of proposed scheme towards this deployment is not a specific case study but to consider a generalized mobility model over dynamic topology in large scale 5G deployment scenario. The target of adopted analytical research methodology is towards optimizing the resources, network management, undertaking complex decision in 5G. The agenda of methodology is also towards enhancing the demands towards quality of service as well as monitoring the network for better management of 5G resources. The proposed scheme implements LSTM owing to multiple rationale: i) in order to facilitate applicability of proposed scheme in real-time traffic scenario, the proposed scheme renders higher smartness in the proposed model with increasing acquisition of mobility-based information, ii) different from other existing learning approaches discussed in Section 2, proposed scheme increases learning accuracy over multiple epoch, and iii) adopted LSTM offers efficient predictive performance for larger dataset with faster fine-tuning of the data compared to conventional models of time-series. The architecture of the proposed scheme is as follow:

According to Fig. 1, the architecture of the proposed scheme is designed considering multiple operational blocks which are briefed as follows:

Deployment Environment: The deployment environment consists of defined path with user-handheld device connected to 5G using various available form of local networks. A user carrying their handheld device travels in a defined set of path where a specific access point is assumed to be installed to cover the set of paths. All the access points are further synced to each other as well as resides within the transmission region of a base station via a gateway node.
5G Implementation: The proposed scheme adheres to principle of RFC 8822 for 5G implementation where the prime functionalities reside in deployed a converged 5G services to access points and gateway node. The packet for 5G based transmission consists of data payload, length, protocol ID, session ID, type, version, and QoS Flow Identifier. The size of session id and protocol id is retained at 16-bits.
Mobility Model: The core role of mobility is played by the user node itself, which is programmed to travelled on a defined path with variable density of other users in uncertain form, with variable speed. The possible direction of mobility in one path is retained as two directions only. The information of the mobility is autonomously captured by the covered access points which retains path information to multiple other nodes residing in range of other access points in sync with the former one. All this aggregated information is further retained over real time server passed via gateway node and base station. This information is also shared among different connected base station.
Predictive Scheme: In order to maintain faster response time without affecting on-going processing over user node, the implementation of predictive scheme is carried within the real-time server connected over cloud. The predictive outcome is further customized based on the query generated by each user node looking for navigational support. The process also facilitates autonomous transmission of updated routes even after the prior predicted routes are generated.

In order to design an effective predictive plan for a traffic management for mobile prone network over 5G, it is essential to consider various constraints associated with the limitation of devices as well as data processing that can be permitted by such nodes. Hence, a proper system design will assist towards formulating better plan towards practical deployment of proposed predictive method.

5.1 Proposed Mobility Model in 5G

According to proposed mobility model, a user node is considered to be capable of connecting to all the forms of network. This is quite feasible where a user node could be considered to be a smart handheld device which can connect itself to different devices as well as networks using Bluetooth, Near Field Communication, Wi-Fi, Infrared sensing, location-based services, cellular services etc. There are multiple forms of connections deployed for this mobility model. An adhoc routing scheme is adopted for connecting all the user’s handheld device which ultimately create a mobility map. The model further constructs a network of path using this mobility map which consists of access point, other user handheld devices, cumulative size of region for mobility of nodes, etc.It is quite essential to ascertain the size of mobility region for each node as well as total area where the mobility is permitted by the user. A condition is constructed where the size of coverage alters (or increase) when the size of overall mobility area is found bigger than demanded size of mobility of an user. The study offers a fixed routes within the boundary of total simulation area for the user to perform its mobility over any direction. Figure 2 highlights the proposed mobility model deployed.

According to the mobility model presented in Fig. 2, the user moves on its assigned path which is characterized by the presence of locally mounted access points. The placement of the access points is carried out in such a way that there is always an established connection among each access point. However, it should be noted that usage of access point is mainly towards serving two requirements viz. i) each access points authenticates the legitimacy of the user and upon successful identification, it generates an authorization key. This key can be used by each node to perform communication with other nodes within the transmission region of that specific access point, and ii) if any user attempts to communicate with other node present in different access point transmission region, than it sends a request to its home access point which further communicates with gateway node to know the foreign access point where the destination node could be found. Upon successful identification, the path information is forwarded by the home access point to the requesting user node. It should be noted that when the user starts moving towards its destination node, this information is autonomously updated so that dynamic movement of destination node does not affect the breakage of link with requestor user node. Apart from this, the study also consider that each access point could possibly use different routing scheme and hence the translation of the heterogeneous routing scheme is carried out by gateway node which is connected with the cell tower. The cell tower is further considered to be connected with other towers using 5G router while it also stores all the mobility map within its region in real-time server connected to cloud. The advantage of this mobility model is that it reduces the dependency of consistent path exploration as all the possibilities of communication vectors keeps on updating in the mobility map seamlessly. Hence, after certain period of time, there is less computational burden for any network device or nodes to look for information for reaching its destination node. Another significant advantage is that it reduces the possibilities of link breakage no matter whatever is the mobility pattern of user node.

5.2 Proposed Predictive Scheme

The proposed scheme implements Long Short-Term Memory (LSTM) in order to carry out predictive analysis. However, proposed scheme doesn’t directly implement LSTM as it is; on the contrary, proposed scheme considers multiple training operations for performing training of LSTM. The advantage of this training scheme is that mechanism assists in amending the weight associated with network of LSTM accordingly to eliminate the errors in the output. It should be noted that nature of the communication data associated with mobility of the user in proposed scheme is potentially of time-series which is needed in order to perform prediction of input and outcome of training. The flexibility of this scheme is also towards opting either single of multiple values of time series from the historical values obtained from mobility map information stored and updated in real-time server. The proposed scheme constructs a communication environment in 5G in order to construct an aggregated data from the users required for training purpose. The complete data is divided into 80% for training and 20% for testing followed by normalization of the input data to fall under probability score of (1, 0). During the initial epoch (i.e., involved iteration series), the parameters associated with bias and weight are arbitrarily assigned. The data is assigned with a known labelled value and is observed by the input layers in the form of an object. The neurons are activated to carry out their job while it transfers its outcome to the next layer till the predictive outcome is reached. Finally, the losses are determined from the difference of actual and predicted outcome using enhanced LSTM. This process is also accompanied by the estimation of the values related to biases and updated weights while the propagation of the fine-tuned outcomes is forwarded to the network layers of neurons. The updated values of the displacement and the weights are used for batch training in the next epoch after successful completion of the prior stage.

The implementation of the proposed study is carried out in MATLAB considering normal 64-bit windows machine. A simulation environment is constructed where an aggregated is generated programmatically. There are two types of data used for investigation. The first type of data is about the request messages being forwarded the user via access point and management with processing in cloud server in compliance of data frame of 5G. The second type of data is that of the predictive outcome which are relayed to the distinct requesting nodes. The proposed model is assessed over 1000 iteration set while 0.1 is the value initially configured for learning rate. The drop rate and period are configured to be 0.2 and 135 respectively based on multiple trials carried out in simulation study. Further, Adam optimizer is used for assembling the learning model followed up by assessing loss analysis that is carried out by mean squared error function. Apart from this, the proposed scheme is compared with multiplemachine learning approaches reported in existing schemes of improving 5G network i.e., Support Vector Machine (SVM), Reinforcement Learning (RL), Federated Learning (FL), Reinforcement Learning with Fuzzy Logic (RL-FuLo), Convolution Neural Network (CNN), and LSTM. The outcome of the comparative analysis is as follows:

Figure 4 highlights the comparative analysis of loss where the observation is carried out for 400, 600, 800, and 1000 iteration. The outcome showcase that proposed scheme offers drastically reduced loss score in contrast to other learning approaches. The outcome shows that SVM and FL approach to offer higher loss score while the performance of RL-FuLo is improved compared to RL. Apart from this, proposed scheme exhibits significant rate of reduction of loss with increasing iteration which is not found in other schemes. It can also be seen that with slightest improvement in LSTM in proposed scheme the performance is improved compared to conventional LSTM for higher iteration.

Figure 5 highlights the outcome of throughput obtained for different learning schemes over multiple set of iterations. The outcome shows that proposed scheme offers better improvement over conventional LSTM. Further, it is noted that throughput performance of CNN over higher iteration (i = 1000) is quite equivalent to that of conventional LSTM. Further, the throughput outcome for RL, FL, and RL-FuLo is somewhat equivalent to each other out of which RL methods shows optimistic increase in data delivery performance. Hence, proposed scheme offers better supportability towards improving data transmission in 5G.

Figure 6 showcase the outcome of delay performance where it can be noted that there is a typical trend of outcome for both proposed (revised LSTM) and conventional LSTM, which is related to consistency trend in increasing delay with increased iteration. CNN is observed with the highest delay score right for maximum set of iterations while the performance of SVM, RL, and FL is nearly equivalent to each other with least significant difference over higher iteration value. However, with iteration of 800, RL is found to offer higher delay compared to SVM and FL. The delay performance of performance RL-FuLo is satisfactorily found to better than CNN, FL, RL and SVM.

According to the observation of the first outcome of loss function, It is noted that proposed scheme offers approximately 35% better performance to average aggregated score of existing learning schemes. The outcome of loss function is dimensionless and is computed by obtaining difference of actual to predicted values. The prime reason for this is as follows: Although, SVM is found to be more efficient towards spaces with higher dimension but it cannot cater up the larger set of data in 5G. FL supports continual learning however it keeps on constructing data on ever local device which are not possible to be updated owing to memory restriction of devices. Hence, they tend to show higher loss in increasing iteration. Further, RL shows minimal loss in initial rounds of iteration; however, increasing round leads to overloading of data saturates the performance of RL. This is slightly improved when RL is integrated with Fuzzy logic (RL-FuLo). Still, this mechanism is highly dependent upon the ruleset being constructed to control it.

Nearly similar reasons attributes towards justifying the outcomes of throughput-based performance. It is computed on the basis of higher score of rates of data transmitted by a unit block of resources into single slot in 5G network. The throughput of proposed scheme is found to be approximately 19% higher compared to existing scheme. CNN is also found to offer better throughput slightly equivalent to LSTM. However, CNN is not recommended for emergency application or some application in 5G that demands mission critical services. It is because of its potential dependencies with large volume of training data. The prime contribution of proposed approach towards improving conventional LSTM can be perhaps seen in delay performance. A closer look into the trends of outcome in every iteration shows that proposed scheme offers better delay performance with change of iteration which is not seen in existing LSTM. For a given set of initializations of an arbitrary weight, the conventional LSTM is found to be highly sensitive and hence it could not showcase the increase in delay as it cannot implement its dropouts. With increasing data, the problem of overfitting in LSTM increases and hence proposed revised LSTM offers better outcome. The outcome exhibited proposed scheme to approximately score 23% of enhanced delay performance in contrast to all the existing scheme. CNN, SVM, and FL offers nearly similar performance for 800 and 1000 iterations. Further, the proposed scheme is assessed with respect to memory consumption and processing time as exhibited as following:

The outcome in Fig. 7 and Fig. 8 showcase that proposed scheme exhibits approximately 37% reduced memory consumption and 41% faster processing time in contrast to existing scheme. Although the proposed scheme is slightly iterative, but its dependency on higher iteration for better fit is quite notable which is one of the computationally cost effective solution for 5G based learning approach deployment compared to existing learning schemes.

The paper presents a simplified and a novel baseline architecture towards facilitating intellectual services towards traffic management with mobility in 5G network. The contribution stated by the proposed study model are as follows: i) a novel deployment environment is constructed where a user is permitted to move on assigned path in two direction in adhoc manner of connectivity with the nearest access point, ii) the scheme deploys RFC 8822 to introduce a converged 5G services considering a standard 5G packet frame, which is yet not reported in any existing methods, iii) a novel mobility modelling is carried out where node density with variable speed plays prime constraint in order to induce the practicality of deployment, iv) a novel enhanced predictive scheme is implemented using LSTM that offers an autonomous predictive updates with dynamic weight and bias management for seamless connectivity, and v) a benchmarked analysis to prove that proposed scheme offers better service delivery performance in contrast to multiple existing learning mechanism.

The future direction of the work will be focused on securing two aspects i.e., i) safeguarding the established link of communication with novel authorization and ii) optimizing the distributed nature of operation towards inclusion of more heterogeneous distributed services in 5G considering use cases.

Acknowledgement	Not applicable
Funding	Not applicable
Conflicts of interest/Competing interests	Not applicable
Availability of data and material	Owing to the nature of the research, due to ethical factor, supporting data is not available
Code availability	Custom Code available and developed by the author
Authors' contributions	A efficient compression modeling proposed based on joint operation of the DWT and Neural Network technique to achieve high compression ratio without much compromising the data quality.

Alasti, H. (2022). Industrial Applications of the Internet of Things and 5G and 6G Networks, ISBN: 9781799892663, 1799892662,
Vairavel, G., Suresh, P., & Saravanakumar, U. (2020). Design Methodologies and Tools for 5G Network Development and Application, IGI Global, ISBN: 9781799846116, 1799846113,
Toskala, A., Holma, H., & Nakamura, T. (2020). 5G Technology-3GPP New Radio, Wiley, ISBN: 9781119236313, 1119236312,
Bhoi, A. K., Barsocchi, P., & Sur, S. N. (2022). V.H. Costa de Albuquerque, 5G IoT and Edge Computing for Smart Healthcare, Elsevier Science, ISBN: 9780323906647, 0323906648,
Nwajana, A. O., & Ihianle, I. K. (2021). Handbook of Research on 5G Networks and Advancements in Computing, Electronics, and Electrical Engineering, IGI Global, ISBN: 9781799869948, 1799869946,
Hussein, H., Abd El-Kader, S. M., Fundamental and Supportive Technologies for 5G Mobile Networks, & Global, I. G. I. (2019). ISBN: 9781799811541, 1799811549,
Sabban, A., & Saban, A. (2020). Wearable Systems and Antennas Technologies for 5G, IOT and Medical Systems, CRC Press, ISBN: 9780367409135, 0367409135,
Kim, D. Y., Jung, M., & Kim, S. (Dec. 2020). An Internet of Vehicles (IoV) Access Gateway Design Considering the Efficiency of the In-Vehicle Ethernet Backbone. Sensors (Basel, Switzerland), 21(1), 98. doi: 10.3390/s21010098.
Tufail, A., Namoun, A., Abi Sen, A. A., Kim, K. H., Alrehaili, A., & Ali, A. (May 2021). Moisture Computing-Based Internet of Vehicles (IoV) Architecture for Smart Cities. Sensors (Basel, Switzerland), 21(11), 3785. doi: 10.3390/s21113785.
Mourtzis, D., Angelopoulos, J., & Panopoulos, N. (Dec. 2021). Smart Manufacturing and Tactile Internet Based on 5G in Industry 4.0: Challenges, Applications and New Trends. Electronics, 10(24), 3175. doi: 10.3390/electronics10243175.
Kourtis, M. A., Siriwardhana, A., Porambage, P., Liyanage, M., Ylianttila, M. (7891). "A Survey on Mobile Augmented Reality With 5G Mobile Edge Computing: Architectures, Applications, and Technical Aspects," in IEEE Communications Surveys & Tutorials, vol. 23, no. 2, pp. 1160–1192, Secondquarter 2021, doi: 10.1109/COMST.2021.3061981.
Y. Siriwardhana, P. Porambage, M. Liyanage and M. Ylianttila, "A Survey on Mobile Augmented Reality With 5G Mobile Edge Computing: Architectures, Applications, and Technical Aspects," in IEEE Communications Surveys & Tutorials, vol. 23, no. 2, pp. 1160-1192, Secondquarter 2021, doi: 10.1109/COMST.2021.3061981.
Chen, S. (May 2022). "Instant Reality: Gaze-Contingent Perceptual Optimization for 3D Virtual Reality Streaming," in IEEE Transactions on Visualization and Computer Graphics, vol. 28, no. 5, pp. 2157–2167, doi: 10.1109/TVCG.2022.3150522.
Thomas, M., Edwards, R. M., & Wang, Z. (2018). "Consideration of IEEE 802.11p and proposed 5G for holograms in vehicular communication," 12th European Conference on Antennas and Propagation (EuCAP 2018), London, UK, pp. 1–5, doi: 10.1049/cp.2018.1245.
Thomas, M., Edwards, R. M., & Wang, Z. (2018). "Consideration of IEEE 802.11p and proposed 5G for holograms in vehicular communication," 12th European Conference on Antennas and Propagation (EuCAP 2018), London, UK, pp. 1–5, doi: 10.1049/cp.2018.1245.
Thomas, M., Edwards, R. M., & Wang, Z. (2018). "Consideration of IEEE 802.11p and proposed 5G for holograms in vehicular communication," 12th European Conference on Antennas and Propagation (EuCAP 2018), London, UK, pp. 1–5, doi: 10.1049/cp.2018.1245.
O’Connell, E., Moore, D., & Newe, T. (2020). “Challenges Associated with Implementing 5G in Manufacturing,” Telecom, vol. 1, no. 1, pp. 48–67, Jun. doi: 10.3390/telecom1010005.
Zikria, Y., Kim, S., Afzal, M., Wang, H., & Rehmani, M. (Oct. 2018). 5G Mobile Services and Scenarios: Challenges and Solutions. Sustainability, 10(10), 3626. doi: 10.3390/su10103626.
Chataut, R., & Akl, R. (May 2020). Massive MIMO Systems for 5G and beyond Networks—Overview, Recent Trends, Challenges, and Future Research Direction. Sensors (Basel, Switzerland), 20(10), 2753. doi: 10.3390/s20102753.
Ahmadi-Assalemi, G., Al-Khateeb, H., Epiphaniou, G., & Maple, C. (2020). “Cyber Resilience and Incident Response in Smart Cities: A Systematic Literature Review,” Smart Cities, vol. 3, no. 3, pp. 894–927, Aug. doi: 10.3390/smartcities3030046.
Rekkas, V. P., Sotiroudis, S., Sarigiannidis, P., Wan, S., Karagiannidis, G. K., & Goudos, S. K. (Nov. 2021). Machine Learning in Beyond 5G/6G Networks—State-of-the-Art and Future Trends. Electronics, 10(22), 2786. doi: 10.3390/electronics10222786.
Domingo, M. C. (Nov. 2021). An Overview of Machine Learning and 5G for People with Disabilities. Sensors (Basel, Switzerland), 21(22), 7572. doi: 10.3390/s21227572.
Malik, S., Khan, M. A., Aadam, H., El-Sayed, J., Khan, & Ullah, O. (2022). “Implanting Intelligence in 5G Mobile Networks—A Practical Approach,” Electronics, vol. 11, no. 23, p. 3933, Nov. doi: 10.3390/electronics11233933.
Imanbayev, A. (Dec. 2022). “Research of Machine Learning Algorithms for the Development of Intrusion Detection Systems in 5G Mobile Networks and Beyond,” Sensors, vol. 22, no. 24, p. 9957, doi: 10.3390/s22249957.
Mazhar, T., et al. (Nov. 2022). The Role of ML, AI and 5G Technology in Smart Energy and Smart Building Management. Electronics, 11(23), 3960. doi: 10.3390/electronics11233960.
Kumudavalli, T., Rayan, & Sandeep, S. C. (2021). "Machine Learning IDS Models for 5G and IoT", Springer-Secure Communication for 5G and IoT Networks, pp73–84, https://doi.org/10.1007/978-3-030-79766-9_5
Wang, Y., & Tian, Z. (2018). Big Data in 5G, Springer-Encyclopedia of Wireless Networks, pp 1–11, DOI: https://doi.org/10.1007/978-3-319-32903-1_58-1
Bega, D., Gramaglia, M., Banchs, A., Sciancalepore, V., & Costa-Pérez, X. (March 2020). "A Machine Learning Approach to 5G Infrastructure Market Optimization,". IEEE Transactions on Mobile Computing, 19(3), 498–512. doi: 10.1109/TMC.2019.2896950.
Koshimizu, T., Gengtian, S., Wang, H., Pan, Z., Liu, J., & Shimamoto, S. (2020). "Multi-Dimensional Affinity Propagation Clustering Applying a Machine Learning in 5G-Cellular V2X," in IEEE Access, vol. 8, pp. 94560–94574, doi: 10.1109/ACCESS.2020.2994132.
Khan Tayyaba, S. (2020). "5G Vehicular Network Resource Management for Improving Radio Access Through Machine Learning," in IEEE Access, vol. 8, pp.6792–6800, doi: 10.1109/ACCESS.2020.2964697.
Strodthoff, N., Göktepe, B., Schierl, T., Hellge, C., & Samek, W. (2019). "Enhanced Machine Learning Techniques for Early HARQ Feedback Prediction in 5G," in IEEE Journal on Selected Areas in Communications, vol. 37, no. 11, pp. 2573–2587, Nov. doi: 10.1109/JSAC.2019.2934001.
Gheisarnejad, M., Mohammadzadeh, A., Farsizadeh, H., & Khooban, M. H. (2022). "Stabilization of 5G Telecom Converter-Based Deep Type-3 Fuzzy Machine Learning Control for Telecom Applications," in IEEE Transactions on Circuits and Systems II: Express Briefs, vol. 69, no. 2, pp. 544–548, Feb. doi: 10.1109/TCSII.2021.3102282.
Polese, M., Jana, R., Kounev, V., Zhang, K., Deb, S., & Zorzi, M. (2021). "Machine Learning at the Edge: A Data-Driven Architecture With Applications to 5G Cellular Networks," in IEEE Transactions on Mobile Computing, vol. 20, no. 12, pp. 3367–3382, 1 Dec. doi: 10.1109/TMC.2020.2999852.
Kucuk, M. F., & Uysal, I. (2022). "Anomaly Detection in Self-Organizing Networks: Conventional Versus Contemporary Machine Learning," in IEEE Access, vol. 10, pp.61744–61752, doi: 10.1109/ACCESS.2022.3182014.
Han, J. Y., Jo, O., & Kim, J. (March 2022). "Exploitation of Channel-Learning for Enhancing 5G Blind Beam Index Detection,". IEEE Transactions on Vehicular Technology, 71(3), 2925–2938. doi: 10.1109/TVT.2021.3140019.
Ramachandran, M., Archana, T., Deepika, V., Kumar, A. A., & Sivalingam, K. M. (2022). "5G Network Management System With Machine Learning Based Analytics," in IEEE Access, vol. 10, pp. 73610–73622, doi: 10.1109/ACCESS.2022.3190372.
Zolotukhin, M., Miraghaei, P., Zhang, D., & Hämäläinen, T. (2022). "On Assessing Vulnerabilities of the 5G Networks to Adversarial Examples," in IEEE Access, vol. 10, pp. 126285–126303, doi: 10.1109/ACCESS.2022.3225921.
Yu, S., Chen, X., Zhou, Z., Gong, X., & Wu, D. (2021). "When Deep Reinforcement Learning Meets Federated Learning: Intelligent Multitimescale Resource Management for Multiaccess Edge Computing in 5G Ultradense Network,". IEEE Internet of Things Journal, 8(4), 2238–2251. doi: 10.1109/JIOT.2020.3026589.
Saha, T., Aaraj, N., & Jha, N. K. (2022). "Machine Learning Assisted Security Analysis of 5G-Network-Connected Systems,". IEEE Transactions on Emerging Topics in Computing, 10(4), 2006–2024. doi: 10.1109/TETC.2022.3147192.
Nguyen, H. T., Van Do, T., & Rotter, C. (2021). "Scaling UPF Instances in 5G/6G Core With Deep Reinforcement Learning," in IEEE Access, vol. 9, pp.165892–165906, doi: 10.1109/ACCESS.2021.3135315.
Hernandez, J. A., Ebrahimzadeh, A., Maier, M., & Larrabeiti, D. (December 2021). "Learning EPON delay models from data: a machine learning approach,". Journal of Optical Communications and Networking, 13(12), 322–330. doi: 10.1364/JOCN.437414.
Sevgican, S., Turan, M., Gökarslan, K., Yilmaz, H. B., & Tugcu, T. (June 2020). "Intelligent network data analytics function in 5G cellular networks using machine learning,". Journal of Communications and Networks, 22(3), 269–280. doi: 10.1109/JCN.2020.000019.
Morgado, A. J., Saghezchi, F. B., Mumtaz, S., Frascolla, V., Rodriguez, J., & Otung, I. (2022). "A Novel Machine Learning-Based Scheme for Spectrum Sharing in Virtualized 5G Networks," in IEEE Transactions on Intelligent Transportation Systems, vol. 23, no. 10, pp. 19691–19703, Oct. doi: 10.1109/TITS.2022.3173153.
Coutinho, R. W. L., & Boukerche, A. (June 2022). "Transfer Learning for Disruptive 5G-Enabled Industrial Internet of Things,". IEEE Transactions on Industrial Informatics, 18, 4000–4007. doi: 10.1109/TII.2021.3107781.
Troia, S., Vanegas, A. F. R., Zorello, L. M. M., & Maier, G. (2022). "Admission Control and Virtual Network Embedding in 5G Networks: A Deep Reinforcement-Learning Approach," in IEEE Access, vol. 10, pp. 15860–15875, doi: 10.1109/ACCESS.2022.3148703.
Tashan, W., Shayea, I., Aldirmaz-Çolak, S., Aziz, O. A., Alhammadi, A., & Daradkeh, Y. I. (2022). "Advanced Mobility Robustness Optimization Models in Future Mobile Networks Based on Machine Learning Solutions," in IEEE Access, vol. 10, pp. 111134–111152, doi: 10.1109/ACCESS.2022.3215684.
Rony, R. I., Lopez-Aguilera, E., & Garcia-Villegas, E. (2021). "Dynamic Spectrum Allocation Following Machine Learning-Based Traffic Predictions in 5G," in IEEE Access, vol. 9, pp.143458–143472, doi: 10.1109/ACCESS.2021.3122331.
Allan, D., Eastlake, D., & Woolley, D. (2021). 5G Wireless Wireline Convergence User Plane Encapsulation (5WE), Internet Engineering Task Force (IETF), RFC 8822,

Authorsdetails.doc

Download PDF

Reviewers agreed at journal
22 Jan, 2023
Reviewers invited by journal
22 Jan, 2023
Editor assigned by journal
20 Jan, 2023
First submitted to journal
19 Jan, 2023

You are reading this latest preprint version

Novel Baseline Computational Predictive Model for Seamless Transmission in 5G

Status:

Version 1

Abstract

Figures

1 Introduction

2 Review Of Literature

3 Core Research Problem

4 Proposed System

4.1 Adopted Methodology

5 System Implementation

5.1 Proposed Mobility Model in 5G

5.2 Proposed Predictive Scheme

6 Result Analysis

7 Discussion

8 Conclusion

Declarations

References

Supplementary Files

Status:

Version 1