Multiple solutions for the coexistence of different radio access technologies operating in the same frequency band have been proposed for 5G and WiFi. Most solutions based on spatial division just consider a small amount of radio access points, one link direction, and/or a single radio access technology. As a consequence, the performance of these solutions on realistic wireless network deployments may be poor and difficult to estimate. This paper investigates the serving of multiple users by multiple radio access technologies with the objective of minimizing the interference among network nodes. This is done by jointly optimizing the beams and link directions as well as the transmission powers, so as to ensure fair and near-optimal throughput allocation over time. For this purpose, a generalized beam-gain model for small-scale antenna arrays is proposed. We evaluate our proposed solution for realistic network scenarios in order to show its effectiveness.

The architecture of 5G networks relies on partial function centralization to reduce operating costs and allow for improved interference management. Nonetheless, partial centralization results in suboptimal operation if the functional split between centralized and distributed functions is not adapted to the time-varying network conditions. Previous work has tackled the problem of selecting the optimal split for fixed conditions, yet it is unclear whether and when dynamic deployments are superior to static ones. In this work, we provide a comprehensive study on the feasibility and profitability of a 5G network featuring a dynamically-adapted functional split. We formulate the selection of this split as a novel optimization problem that combines performance and operating cost into revenue. Moreover, we base on a dedicated cost model to propose multiple adaptation strategies and evaluate them under a wide range of network designs and conditions. Finally, we show that a dynamic functional split may lead to substantially cost reductions (or revenue increases) with respect to static configurations.

Centralization in 5G radio access networks brings two main benefits: reducing cost and improving performance. Although an ideal, fully-centralized architecture would provide minimum cost and maximum performance, actual deployments cannot simultaneously optimize both. Previous research focuses on how to select the functional split of a 5G network to either maximize performance or minimize cost on partially-centralized architectures, without exploring which approach is the most appropriate. In this work, we investigate the trade-off between cost and performance of both approaches, in order to figure out which one is more adequate for real network operators. We provide a comprehensive study under a wide range of network conditions and show that, in general, a performance-maximizing approach is more likely to produce a higher net revenue.

Communication networks are evolving towards a more adaptive and reconfigurable nature due to the evergrowing demands they face. A framework for measuring network flexibility has been proposed recently, but the cost of rendering communication networks more flexible has not yet been mathematically modeled. As new technologies such as software-defined networking (SDN), network function virtualization (NFV), or network virtualization (NV) emerge to provide network flexibility, a way to estimate and compare the cost of different implementation options is needed. In this paper, we present a comprehensive model of the cost of a flexible network that takes into account its transient and stationary phases. This allows network researchers and operators to not only qualitatively argue about their new flexible network solutions, but also to analyze their cost for the first time in a quantitative way.

In order to deliver the high data rates promised for 5G networks, mobile base stations need to be deployed in dense layouts. This results in increased inter-cell interference, which can be mitigated by leveraging centralized architectures in radio access networks. Nonetheless, centralizing all the processing requires prohibitively high link capacities for the fronthaul network connecting centralized and distributed units. In contrast, a static, partially-centralized architecture yields poor performance as it fails to adapt to instantaneous interference situations. In this work, we show that a dynamically centralized architecture enables drastic interference reductions even when using a very limited fronthaul network. We propose multiple algorithms to find the optimal centralization option and evaluate their performance on operator-grade hardware. In addition, owing to the dynamicity of the problem being solved, we provide a framework to decide on the best algorithm based on the trade-off between performance, cost, and adaptation time.

The architecture of the radio access network (RAN) in 5G features a functional split between centralized and distributed units, which can be leveraged to reduce inter-cell interference. Recent work proposes to dynamically adapt this split in accordance with the instantaneous interference situation experienced by all users. However, it is unclear whether performing this flexible adaptation is actually feasible, since the interference situation changes continuously as users move. In this work, we investigate the impact of mobility on the problem of dynamically selecting the optimal functional split. We employ a mobility simulator based on real street layouts and trace-derived traffic patterns to generate continuously varying interference situations. Then, we analyze how frequently the optimal functional split changes and how much the performance of previous splits differs from the new, optimal one. The results allow us to estimate the time required for a viable flexible functional split adaptation.

In the field of networking research, increased flexibility of new system architecture proposals, protocols, or algorithms is often stated to be a competitive advantage over its existing counterparts. However, this advantage is usually claimed only on an argumentative level and neither formally supported nor thoroughly investigated due to the lack of a unified flexibility framework. As we will show in this paper, the flexibility achieved by a system implementation can be measured, which consequently can be used to make different networking solutions quantitatively comparable with each other. The idea behind our mathematical model is to relate network flexibility to the achievable subset of the set of all possible demand changes, and to use measure theory to quantify it. As increased flexibility might come with additional system complexity and cost, our framework provides a cost model which measures how expensive it is to operate a flexible system. The introduced flexibility framework contains different normalization strategies to provide intuitive meaning to the network flexibility value as well, and also provides guidelines for generating demand changes with (non-)uniform demand utilities. Finally, our network flexibility framework is applied on two different use-cases, and the benefits of a quantitative flexibility analysis compared to pure intuitive arguments are demonstrated.

Along with many other novel features, the fifth generation of mobile networks (5G) aims at highly flexible and dynamic network management, as well as reduced cost for operators. In order to enable both features, rapid and efficient adaptation to environmental changes is needed. This requires a complete knowledge of the characteristics of the user traffic at all time scales, but state-of-the-art research clearly differentiates between large-scale and small-scale traffic behavior. In this work, we propose a traffic model that connects large-scale and small-scale phenomena. We show that the standard small-scale models may produce inaccurate results in case of network congestion. We propose a strategy to mitigate this problem and evaluate it through simulations.

The 3rd Generation Partnership Project (3GPP) proposes a centralized architecture for the 5G radio access network (RAN) in order to reduce costs and mitigate inter-cell interference, which helps to increase user data rates. However, the limited capacity of current fronthaul networks renders it impossible for many RANs to be fully centralized. Instead, the operators can opt for a partially centralized architecture, in which only some of the functions of the RAN's processing chain are centralized. Previous work has tackled the optimal selection of these functions in a static or semi- static manner. In this paper, we present a 5G RAN that is able to dynamically adapt the subset of centralized functions to maximize data rates at runtime. We analyze the dynamics of a dense 5G RAN to derive a maximum convergence time for the selection algorithms and show that a dynamic functional split significantly improves data rates with respect to statically centralized solutions.

In the 5G radio access network (RAN), the functions of the next-generation eNodeB (gNodeB) are split into a centralized and a distributed unit. Depending on how the split is performed, the amount of traffic generated between the units can be too high to be supported by the current infrastructure. Therefore, a careful characterization of this traffic is needed for every split option. Among the wide array of options, the MAC-PHY split proves to be both promising, as a balanced trade-off between centralization and decentralization, and difficult to characterize, due to the high amount of low level interactions between MAC and PHY layers. Indeed, the MAC-PHY split is frequently considered in the literature as a possible implementation option for the 5G RAN, yet there is no detailed study about the capacity it requires. This paper remedies that by offering a comprehensive analysis of the downlink traffic of a MAC-PHY split for 5G networks. This analysis is backed with both simulative data and measurements from a physical implementation.

The performance of cloud computing depends heavily on the networking infrastructure, which carries the data traffic between the VMs, as well as the essential control and management traffic. Meanwhile, networks in data centers are becoming softwarized with the Software Defined Networking (SDN) paradigm to fuel their programmability and flexibility. However, since SDN switches need to contact the controller frequently for their configuration and flow rules setup, the overall data traffic forwarding performance can suffer, if the control and management plane gets congested. In order to proactively avoid such congestion, we need to study the involved traffic in detail. In this work, we perform an elaborate traffic evaluation on a real OpenStack cloud deployment and evaluate the types of exchanged messages in the control and management plane and their respective traffic volume. The evaluation reveals that the control and management traffic scales with the number of VMs and the VM-related events, and therefore provides guidelines for planning and operating the cloud networking infrastructure.

The problem of wireless M2M communication is twofold: the reliability aspect and the scalability aspect. The solution to this problem demands a delay constrained random access protocol. To this end, we propose admission control based traffic-agnostic delay-constrained random access (AC/DC-RA) protocol. Our main contribution is enabling the stochastic delay constraints agnostic to the traffic, such that the stochastic delay constraint is valid with respect to a varying number of arrivals. We achieve this with an admission control decision that uses a novel collision estimation algorithm for an active number of arrivals per contention resource. We use an adaptive contention resolution algorithm to react to the varying number of arrivals. Using these tools, the admission control solves the stability problem. We show with simulations that the AC/DC-RA provide stochastic delay constrained random access in a traffic agnostic way to sustain a stable performance against Poisson and Beta arrivals without any modification to the protocol.

The design of the 5G next-generation radio access network (NG-RAN) proposes the division of the next- generation eNodeB (gNB) into centralized and distributed units. Centralization should facilitate coordination of RAN functions between gNBs, but the actual benefits of it are still unclear. In this paper, we provide a study of the feasibility and benefits of coordinated downlink scheduling. We first analyze the time constraints that a coordinated scheduler has to face from a theoretical perspective, and we back them with an experimental proof-of-concept. Then, we present a lightweight scheme for coordinated link adaptation that matches the previous constraints. We show that coordination is indeed feasible with state- of-the-art technology, although very limited by time constraints. Finally, we show the results of our experimental testbed, which successfully implemented the described coordination scheme under the predicted constraints.