Unmanned Aerial Vehicles (UAVs) are expected to be an important component for supporting high data rate transmissions in network slice enabled 5G networks. UAVs provide the opportunity to realize the concept of aerial base stations which can help in boosting coverage, spectral efficiency, and overall quality of experience of users. Network slicing logically divides the network resources including bandwidth, and provides diversified service requirements for different application scenarios and enhances network performance. However, slicing and allocation of available bandwidth to various users act as a challenge in achieving desired network performance. Additionally, energy constraints leads to limited hover time at UAVs which introduces further challenges in serving users' requests. In this work, we propose optimal rate and distance based algorithms which allocate bandwidth slices to users with the objectives of minimizing the hover time and maximizing the number of users which are served, while considering limitations pertaining to available bandwidth, hover time of UAVs, and quality of service to the users. Simulation results show that our proposed algorithms result in significant improvement over state-of-the art bandwidth allocation techniques.

In the softwarized fifth generation (5G) networks, authentication protocols are deployed by both the mobile network operators and the third-party service providers to enable secure slice formation at the network side and user validation at the service provider side, respectively. However, the usage of static key-based authentication protocols in 5G networks gives rise to two major problems: (i) network slice topology learning attack during the formation of slice and (ii) exposure of users’ service access behavior to the third-party service providers and device-to-device communication peers. In this article, we propose group anonymous (i) privacy preserving mutual authentication (PPMA) and (ii) privacy preserving one-way authentication (PPOA) protocols to address the aforementioned problems, respectively. We also prove the security of the PPMA and PPOA protocols against the chosen ciphertext attack in the random oracle model. The PPMA and PPOA protocols are implemented using the JPBC library and the computation overhead is measured. For 1024 bits discrete logarithm security, the computation overhead of the PPMA and PPOA protocols is respectively 48.54 and 68.65 percent lower than an existing privacy preserving authentication protocol on an Intel processor. The PPMA and PPOA protocols are generic and can be applied in peer-to-peer authentication scenarios where group anonymity and privacy preservation are considered.

Cloud Radio Access Network (C-RAN) is a crucial component in 5G network deployments. The C-RAN technology will reduce operating costs, improve energy efficiency, and support enhanced capacity and coverage by taking advantage of centralized processing. However, C-RAN requires strict latency bounds on the fronthaul network that connects the distributed radio units and the central processing units. Instead of deploying dedicated fronthaul transport networks, the reuse of existing backhaul transport networks to carry fronthaul data needs to be leveraged. In this paper, we study the end-to-end delay bounds for a mixed fronthaul and backhaul 5G network. Deterministic network calculus is used to derive worst-case delay bounds in a mixed transport network, and this is verified using a C-RAN experimental testbed with real traffic characteristics. The impact on end-to-end delay due to the effect of multiplexing fronthaul and diurnal backhaul traffic in business and residential regions is studied analytically. We also propose a framework to determine the choice of enabling the appropriate fronthaul split at the distributed radio units and the central processing units to meet the latency requirements. The proposed network calculus based framework can be utilized by network engineers while designing the future 5G transport network deployments.

Mobile Network Operators perceive Spectrum Slicing and Sharing as an efficient and viable option to meet the growing mobile user data demands in the upcoming 5G network deployments. Efficient spectrum sharing requires optimized spectrum partitioning, i.e., slicing of the spectrum for sharing among operators, to improve spectrum utilization, offer services with high quality of experience, and increase their revenues. This paper proposes Spectrum Slicing Coalition Formation (SSCF) protocol, a novel solution to address effective spectrum slicing required to meet the dynamic data demands at every 5G base station. The base stations' cooperation is formulated as a coalition formation game with overlapping coalitions, where a base station's contribution is not restricted to a single coalition, instead a base station participates intelligently in multiple coalitions. The SSCF protocol enables a base station to dynamically slice and share its allocated spectrum to service its load based spectral requirements, a key requirement for effective radio access network softwarization. The study explores operator base stations' options to determine with whom, when, and how much to slice and share the spectrum and the payoffs. Extensive simulations using real-world operator base station deployments highlight that SSCF solution brings in operational efficiencies, improved users' quality of experience, and revenue maximization.

Cloud Radio Access Network (C-RAN) technology is a key enabler for 5G network deployments. It brings in the benefits of centralized processing to support dense deployments and to reduce the operating costs. The flip side of C-RAN is it places stringent latency and bandwidth requirement on the fronthaul network connecting radio units and central units. In this work, we propose an efficient topology-aware synchronized packet generation scheme using a centralized SDN controller to reduce the overall network congestion. The study also provides an insight into packet-based SDN-enabled C-RAN behavior modeled as a network of queues and is analyzed by an extensive simulation using the traffic characteristics captured from a C-RAN testbed implementation. The analysis provides useful insights for network designers in enabling future C-RAN deployments.

Slicing of a 5G network by creating virtualized instances of network functions facilitates the support of different service types with varying requirements. The management and orchestration layer identifies the components in the virtualization infrastructure to form an end-to-end slice for an intended service type. The key security challenges for the softwarized 5G networks are, (i) ensuring availability of a centralized controller/orchestrator, (ii) association between legitimate network slice components, and (iii) network slice isolation. To address these challenges, in this paper, we propose a novel implicit mutual authentication and key establishment with group anonymity protocol using proxy re-encryption on elliptic curve. The protocol provides (i) controller independent distributed association between components of a network slice, (ii) implicit authentication between network slice components to allow secure association, (iii) secure key establishment between component pairs for secure slice isolation, and (iv) service group anonymity. The proposed protocol's robustness is validated with necessary security analysis. The computation and bandwidth overheads of the proposed protocol are compared with that of the certificate based protocol, and our proposed protocol has 9.52% less computation overhead and 13.64% less bandwidth overhead for Type A1 pairing.

Internet of Things (IoT) in a real world scenario involves collecting data by sensor devices and communicating the sensed data, through gateway devices, to the IoT Cloud (or the Internet) using direct or multi-hop communication. A gateway typically supports multiple wireless sensor networks (WSNs) and each WSN might use a different technology for communication. Hence gateways are designed to be heterogeneous, having necessary hardware support only for those technologies corresponding to their associated WSNs. Due to their heterogeneity, the technology specific channel access mechanisms need not provide fairness in data communication at each gateway. In this paper, we propose two centralized techniques that involve an efficient link selection per technology and time based scheduling among gateways. This improves fairness in data transmission for every gateway, resulting in an increased data transmission. The first technique uses a multinomial probability based link selection thereby reducing contention within each technology link for a given network topology. The second technique incorporates adaptation to the changing traffic patterns in the network. Our extensive simulation results show that the proposed techniques outperform default technology specific channel access mechanism in IoT networks in terms of total data transmitted, packet delivery ratio, and fairness.

As the mobile data traffic continues to grow rapidly, spectrum availability is a major concern. Furthermore, the fixed spectrum allocation to the mobile network operators (MNOs) and lack of free spectrum result in poor user quality of experience (QoE) and inefficient spectral resource utilization. Inter-operator spectrum sharing is a solution to overcome the spectral shortage, and this is achieved either by an MNO leasing other MNO's spectrum or by sharing a common pool of MNOs' spectrum. Although both approaches bring benefits, they have further challenges. We propose a novel spectrum sharing paradigm called inter-operator proximal spectrum sharing (IOPSS), where a base station (BS) intelligently offloads users to the neighboring BSs based on spectral proximity to enhance the users' QoE and spectral resource utilization. Users requesting high service rates can be served by using carrier aggregation. We demonstrate the IOPSS's benefits using a continuous-time Markov chain-based analytical model of a BS. A generic IOPSS Q-learning framework (IOPSS-QLF) for a BS to dynamically determine its load-based spectral needs and efficiently share its spectrum resulting in a self-organizing spectrally efficient network of BSs is proposed. The effectiveness of IOPSS-QLF is demonstrated using extensive simulations.

With the evolution of various new types of application services for mobile devices, cellular operators have started providing multiple subscription plans to the mobile users. The plan subscribed determines the Quality of Service (QoS) to be provided to the user, and operators distinguish users as priority users and non-priority users, accordingly. To ensure better QoS to the priority users, necessary resources must be reserved at the base station. This demands analyzing the network traffic to predict future traffic pattern. This paper pioneers the use of machine learning methods to forecast mobile user traffic pattern for providing better QoS to the priority users. We analyze two different supervised learning methods, Naive Bayes Classifier and Logistic Regression, used for prediction of probable times when a priority user would be active. The prediction results are applied to the user scheduling strategies for efficient bandwidth management, thus improving system capacity and reduce blocking. Simulations on multiple real-life datasets validate the model and predict the mobile user pattern with very high accuracy, along with significant reduction in priority user service blocking ratio and improvement in their capacity.

Effective ways to address spectrum shortage by sharing cellular spectrum among the operators are constantly explored. Inter-Operator spectrum sharing, where operators dynamically share their licensed spectrum, when the network capacity requirement varies based on user access is an option to address this issue. The emerging licensed Co-Primary spectrum sharing, a commonly shared spectrum band used among a set of operators is an alternative option. Both these options require the supporting network infrastructure to adapt dynamically to the different bandwidth processing requirements, and this is a challenge. Adapting the proximity based access prevalent in the financial networks, which enhances customer's Quality of Experience (QoE), this study explores the benefits of exploiting the proximal based access in cellular networks. Proximal Spectrum Access (PSA) enables a user to access the nearest base station irrespective of his/her service provider association, which enhances the user's QoE and lessens the strain on network resources. With extensive simulations based on real life multiple operator deployment scenarios, this study validates the deployment benefits to operators as well as end user QoE enhancements.

Densely deployed small cell base stations are a key factor to enable 5G networks. Multicore System-on-a-Chip (MCSoC) solutions supporting small cell LTE base station functionality are commercially available. An MCSoC has to be integrated with necessary software components leveraging specific accelerators for creating a complete LTE base station. System architects have a tough challenge in determining whether a small cell MCSoC would meet their intended small cell base station's feature, functionality and performance demands. This paper discusses the application of modular performance analysis (MPA) using Real-Time Calculus (RTC) on an MCSoC-based LTE base station implementation. Our study validates its applicability as well as its ability to identify new designs that will improve the system performance. This study is first in its kind to propose an analytical framework for analyzing an LTE base station's system design and performance.