The deployment of 5G networks, which are characterized by high-frequency bands, dense cell deployments, and users with diverse application demands, presents significant challenges for efficient mobility management. Frequent handovers lead to transmission interruptions and reduced network capacity, affecting both the user experience and overall network performance. The current standard handover algorithm selects target base stations solely based on signal strength, without considering the availability of resources, which can result in overloaded cells. To address these challenges, in this paper, we formulate a multi-objective optimization problem that reliably captures the network support for dual connectivity. The objective is to maximize network throughput (of interest to the operator), and to minimize service interruptions (of interest to users). Given the computational complexity of the optimization problem, we develop a Deep Reinforcement Learning (DRL)-based algorithm to perform user-to-BS assignment and wireless resource allocation. We evaluate our approach through realistic simulations in two scenarios: an indoor factory floor that uses Frequency Range (FR) 2 and a two-tier outdoor network that operates in both FR1 and FR2. The proposed algorithm demonstrates significant improvements over state-of-the-art methods, achieving better sum throughput and zero handover rates while operating within 10% of the optimum. Additionally, our algorithm ensures a 100% user rate satisfaction. Instead of performing handovers, it utilizes dual connectivity initiations and keeps their rate 20−30% lower than the handover rate.

Within the research for the sixth generation of mobile networks, so-called in-X subnetworks (SNs) were proposed to support services with extreme performance requirements in geographically confined areas. To facilitate efficient spectrum usage and reliable connectivity within in-X SNs, adequate dynamic frequency planning methods for managing interference must be developed. This work provides background on in-X SNs and reviews related work on frequency planning and interference management. The problem of minimizing SN reconfigurations is identified as a key objective, while it is not addressed in the literature to date. Thus, the trade-off between minimizing subband usage and reducing SN reconfigurations is analyzed using both an already existing basic and a newly proposed advanced interference model. Preliminary results highlight how interference modeling and knowledge of future interference scenarios affect optimization outcomes, and future research directions toward dynamic frequency planning for in-X SNs in real 6G systems are discussed.

5G makes it possible to reach to very low latency and very high reliability for different use cases. However, it is not clear how mobility procedures would impact the latency and the reliability of these use cases. Therefore, in this paper, we investigate how a packet flows through 5G RAN Protocol Stack and precisely how handovers affect that flow. Later, we present simulation studies conducted under three different scenarios, Future Railway Mobile Communications System Scenario (a high speed train use case that requires very high reliability), a highway scenario for C-V2X communications and an FR2 Urban Scenario for XR use cases. Our findings indicate that optimizing for handover interruption time can improve reliability up to 0.0094% in the FRMCS scenario, up to 0.045% in the Highway scenario, and up to 0.0071% in the FR2 scenario, underlining the importance of mobility procedures for use cases that require very high reliability, up to 99.9999%.

The rapid development of 5G brought advanced capabilities to support low-latency and bandwidth-hungry applications, including edge computing capabilities. However, service orchestration on the network edge is characterized by the scarcity of available resources in the edge infrastructure, emphasizing the need for optimal resource allocation. In this work, we explore a multi-objective optimization approach to resource allocation in 5G edge networks that focuses on the deployment of edge applications and User Plane Functions (UPFs) to maximize operator’s profit with minimum energy consumption. Due to the computationally expensive nature of the optimization formulation, we develop a solution that relies on the use of genetic algorithms, specifically a non-dominated sorting genetic algorithm (NSGA-II)-based multi-objective approach. We compare the performance obtained with our approach against the optimal solution, where the latter can be obtained only for small instances, and another benchmark heuristic. We do this for three different topologies under varying demand levels. Results show that our approach yields solid performance, outperforming the benchmark by at least 25%.

Massive Machine-Type Communications are one of the service types supported in 5G. They are characterized by massiveness and sporadic traffic, as well as to be energy efficient. While the scattered nature of the traffic occurrence does not pose a serious burden on efficient network planning, the requirement to serve a massive number of devices and to be energy-efficient certainly does. Moreover, besides the successful transmission/reception, these data need to be processed too. With limited resources on both the Radio Access Network part (responsible for communication) and edge cloud part (responsible for processing), as well as with the competition among a large number of users with this traffic in the cellular network, we need to address the problem of admission control, so that the network can successfully serve all the users that were admitted. To that end, in this paper we model the behavior of the system using a queueing network and perform the analysis that leads to admission policies for mMTC traffic with computation requirements. We do this both for homogeneous and heterogeneous users. Using data from a 5G trace, we validate our analytical results and provide further insights. Results show that the number of admitted users almost completely depends on the traffic pattern and that the entity with the lower capacity determines the number of admitted users.

Within the development of 6G, so-called subnetworks were proposed to serve special use cases like intra-vehicle sensor-actuator communication, robot control in industrial environments, or health monitoring. These use cases are characterized by extreme communication demands between the devices served by a single subnetwork. Moreover, the subnetworks will be densely deployed, with mobile and autonomous vehicles carrying the subnetwork Access Points (APs). These properties necessitate novel approaches for frequency planning in order to enable reliable communication within all subnetworks and efficient resource usage. In this context, the problem of dynamic frequency planning for mobile 6G in-X subnetworks is investigated in this paper. To this end, a multi-objective optimization problem with the objectives of minimizing frequency subband usage and subnetwork reconfigurations leveraging knowledge about future interference scenarios is formulated. Afterward, the problem is shown to be NP-hard, and two heuristic algorithms are developed. Using realistic vehicular movement data from simulations, results show that the heuristics outperform a State-of-the-Art (SotA) benchmark. Moreover, the value of knowledge about future interference scenarios is shown. Reconfigurations can be reduced by 18.91% when prioritizing subband usage and even by 33.02% when prioritizing reconfigurations if interference scenarios are known for three time steps instead of one.

Medical applications, such as telemedicine or smart operation rooms, place stringent requirements on the underlying network architecture. 6G as the next-generation communication standard currently in research promises to satisfy the needs of such applications by utilizing advances in technology and networking concepts. One crucial concept for medical applications is the capability of using computing resources within the network. By placing the applications on such processing nodes in different locations within the Radio Access Network (RAN), the performance metrics of a medical application, such as latency, throughput, and availability can be optimized. However, problems arise when the available processing capabilities are not sufficient for all requested medical applications. In this paper, we formulate an Integer Linear Program (ILP) to address the problem of processing medical applications within the network when the processing capabilities are not sufficient. We consider the priority and different service levels of application functions and aim to place as many applications as possible with the best possible service quality. Additionally, we take into account that some applications must run in the network even if their priority is low. Furthermore, we propose a heuristic in order to obtain a good solution quickly. The evaluation of our solution and comparison to existing approaches shows an increase of accepted demands in the network by up to 35%.

Mobile users frequently face significant interruptions in transmission and reception during handovers from one Base Station (BS) to another, resulting in latencies that are incompatible with the stringent requirements of Ultra-Reliable Low Latency Communications (URLLC). To address this, 3GPP introduced a novel handover procedure, called Layer 1/Layer 2 Triggered Mobility (LTM), in Release 18. LTM uses lower level signaling to respond quicker to mobility events, bypassing the reconfiguration of higher layers while keeping modifications to the lower layers at a minimal level. This drastically reduces service interruptions during handovers, making them practically negligible. However, since LTM uses more frequent L1 measurements, it has a higher handover and ping-pong handover rates, as well as signaling overhead. In this work, we propose to incorporate future channel predictions in LTM to perform cell preparations and handover decisions with the goal of reducing signaling overhead and resource reservation. We focus on a controlled indoor scenario, where future user channel predictions are possible with a high accuracy. Our proactive algorithm reduces the cell preparation rate by 76 % and the handover rate by 72 %, without compromising the network sum throughput. Moreover, the resource reservation time at the target BS is reduced to nearly 0 ms.

With the development of 5G, network slicing was proposed to enable service provisioning for diverse sets of Ultra-Reliable Low-Latency Communications (URLLC), enhanced Mobile Broadband (eMBB), and massive Machine-Type Communications (mMTC) users which are characterized by different Quality of Service (QoS) demands. Within network slicing, Radio Access Network (RAN) slicing plays a central role for efficient resource management. In addition, user admission control poses a major challenge. In this context, the problem of joint slice dimensioning and user admission control is investigated in this paper. To this end, an optimization problem based on symbol-level resource allocation with the objective of maximizing an operator's revenue while fulfilling the traffic requirements of all users is formulated. Afterward, the optimization problem is reduced to a knapsack problem and integrated into a Long-Term Revenue Maximization (LTRM) algorithm. Using data from real-world 5G measurements, the efficiency of the LTRM algorithm is verified, and the impact of various resource granularities in the time domain (symbol vs. slot) and frequency domain (varying Resource Block Group (RBG) sizes) is investigated. The revenue gain of the proposed joint algorithm over a sequential slice dimensioning and user admission control scheme is 24%, while symbol-level resource allocation offers at least 13% gain over a slot-based allocation for specific slices.

Navigation and trajectorial estimation of maritime vessels are contingent upon the context of positional accuracy. Even the smallest deviations in the estimation of a given vessel may result in detrimental consequences in terms of economic and ecologic quotients. To ensure an agile and precise environment for maritime vessel positional estimation, preexisting marine radar technologies can be utilized in a way that ensures a higher level of precision compared to GNSS-based identification and positioning. In this paper, we present a positional optimization for radar-based vessel navigation systems that utilize the installment of vessel detection sensors. The main objective of this research is to employ as fewer sensors as possible while preserving the attainable error threshold for positioning that is defined by International Maritime Organization (IMO). Our approach leads most of the time to a positioning error of up to 5m along shorelines and rivers and up to 50m along open coastal regions.

Providing a high-quality real-time video streaming experience to mobile users is one of the biggest challenges in cellular networks. This is due to the need of these services for high rates with low variability, i.e., stable throughput, which is not easily accomplished given the competition among (an ever-increasing number of) users for limited network resources and the high variability of their channel conditions. A way to improve the user experience is by exploiting users’ buffers and the ability to provide a constant data rate to everyone, as one of the initially envisioned features of 5G networks. However, it was already shown that the latter is not very efficient, neither in terms of the achievable data rates nor in terms of the amount of resources left unused. In this paper, we provide a theoretical-analysis framework for resource allocation in 5G networks that leads to an improved user experience when watching live video while providing a constant video resolution at almost all times. We do this by solving four problems, in which the objectives are to provide the highest achievable video resolution to all single-class and multi-class users, and to maximize the number of users that experience a given video resolution. The analysis is validated by simulations that are run on publicly-available traces. We also compare the performance of our approach against other techniques for different Quality of Experience metrics. Results show that performance can be improved by at least 15% with our approach compared to state of the art.

Mobility management in 5G is challenging due to the usage of high frequencies and dense cell deployments. As a result, users experience frequent handovers that cause an interruption in transmission/reception and diminish network capacity. In the common handover algorithm, the target Base Station (BS) is selected based solely on the signal strength, while the available resources are not considered, leading to overloaded cells, especially for macro cells with large coverage. Advanced handover techniques are needed in 5G to perform smooth network operation. In this paper, we formulate an optimization problem, whose goal is to provide α-fairness in data rates among users and to reduce handovers. To accomplish that, we jointly perform user assignment and resource allocation while accounting for the interruption due to handovers. This is an integer nonlinear program and, by relaxing it, an upper bound is obtained. Further, because of the time complexity of the original problem, we propose a Deep Reinforcement Learning (DRL)-based algorithm, which finds near-optimal user-to-BS assignments and the amount of resources that should be allocated to a user. Our approach outperforms considerably state of the art in terms of fairness and handover rate while being within at most 12% of the optimum in most cases.

Edge computing has become a very popular service that enables mobile devices to run complex tasks with the help of network-based computing resources. However, edge clouds are often resource-constrained, which makes resource allocation a challenging issue. In addition, edge cloud servers must make allocation decisions with only limited information available, since the arrival of future client tasks might be impossible to predict, and the states and behavior of neighboring servers might be obscured. We focus on a distributed resource allocation method in which servers operate independently and do not communicate with each other, but interact with clients (tasks) to make allocation decisions. We follow a two-round bidding approach to assign tasks to edge cloud servers, and servers are allowed to preempt previous tasks to allocate more useful ones. We evaluate the performance of our system using realistic simulations and real-world trace data from a high-performance computing cluster. Results show that our heuristic improves system-wide performance by 20-25% over previous work when accounting for the time taken by each approach. In this way, an ideal trade-off between performance and speed is achieved.

With the rapid increase in the amount of exchanged traffic over cellular networks, stemming partly from computation-intensive tasks, and the highly mobile nature of the users, mobility management exhibits considerable challenges in next-generation cellular networks. A way to alleviate these problems is by using Software-defined Radio Access Networks (SD-RAN), where a centralized controller with a complete overview of the network topology (distribution of users across base stations and their channel conditions) can make decisions on the user assignment and resource allocation. To that end, in this paper, we formulate an optimization problem with the objective of maximizing the network utility, where computation-intensive tasks are sent from the users to edge clouds, taking into account the communication constraints (uplink and downlink bandwidth) as well as the finite storage and processing capabilities of edge clouds. Moreover, we provide a user rate guarantee to satisfy an additional application for all users. The problem is NP-hard, therefore, we propose to use Deep Reinforcement Learning (DRL) to solve it. Extensive realistic simulations show that our approach is close to the optimal solution, where the latter is obtained using a solver, while outperforming a benchmark by up to 65%.

Through using edge computing services, mobile devices can run complex tasks with the help of network-based computing resources. However, servers in the edge cloud are not only constrained to limited resources, but also must make allocation decisions with only limited information available. The clients requesting computing resources may also have limited information about the servers available to them. We focus on a distributed resource allocation method in which servers operate independently and do not communicate with each other, but interact with clients to make allocation decisions for those clients’ tasks. We follow a two-round bidding approach to assign tasks to edge cloud servers. Servers may choose to preempt previous tasks to allocate more useful ones, and clients may choose to track the outcomes of their tasks to inform their future decisions. Results show that user learning improves system performance by 50-80% when servers are heterogeneous in pricing aggressiveness.

GSM-R (GSM for Railways) is a 2$G$-based standardized ground-to-train communications system that enabled interoperability across different countries. However, as a 2G-based system, it is nearing its lifetime and therefore, it will be replaced with 5G-based Future Railway Mobile Communications System (FRMCS). FRMCS is expected to bring in new use cases that demand low latency and high reliability. However, from a mobility perspective, it is not clear how the low latency and high reliability will be achieved. This paper investigates the effect of handover procedure on latency and reliability and analyzes which use cases of FRMCS can be satisfied using baseline handover. We also sweep through different handover parameter configurations and analyze their effect on mobility performance. Then, we analyze the effect of mobility performance on packet latency and reliability. Our results show that, with baseline handover, Standard Data Communications Scenario is met and optimizing for baseline handover performance can reduce latency by up to 18.5%, indicating that optimizing for mobility performance is crucial in FRMCS.

Crowdsourcing data collection from a network of mobile devices is useful in various applications. Mobile devices store a large amount of visual data that can aid in different application scenarios. Trained Convolutional Neural Networks (CNNs) can be deployed on mobile devices to be used in searching for objects of interest. Querying for novel objects, for which models have not been trained yet, presents some unique challenges. When novel objects are queried, new models must be trained and distributed to all edge devices. In this paper, we propose an efficient method and a system, called EDIR, which enables answering these queries while taking into account the bandwidth limitations encountered in wireless networks, as well as the limited energy and computational power on mobile devices. Through extensive experimentation, we show that using distance-based classifiers, specifically those relying on the Cosine distance, leads to more efficient utilization of network resources by reducing the number of false positives. We perform analysis that enables the requester to tune the parameters of interest before issuing the query, and validate our theoretical results. EDIR reduces the amount of transferred data by more than 45% compared to other approaches while simultaneously achieving a good F1 score.

While LTE networks are known to provide relatively high data rates, reaching values as high as tens of Mbps, these rates exhibit considerable variability over time. The rate variability hurts especially the performance of applications and services that require stable data rates, such as real-time video streaming, online gaming, virtual reality, augmented reality, etc. 5G emerged as a solution to this as well as to many other problems. However, it has been shown that strict constant data rates come at the cost of underutilized network resources, resulting in inefficient operation of cellular networks. Therefore, a tradeoff between the data rate stability, important to cellular users, and the efficient utilization of resources, important to network operators, needs to be taken into account. To that end, in this paper, we consider the problem of allocating all the network resources to cellular users in such a way that it provides as high a data rate as possible to all users while limiting the rate variation within tight bounds. We do this for different scenarios in terms of the user activity, user type, and the nature of the policy. Firstly, we consider the case of static allocation policy, irrespective of channel conditions, for users that are always active. Then, for these same users, we look at the case when resources are allocated dynamically over time. Secondly, we consider static and dynamic policies for users that are only intermittently active. Thirdly, we consider the case with users having different Service Level Agreements (SLAs) with the cellular operator. Furthermore, we run extensive simulations with input parameters from real traces. Results show that allocating the resources dynamically improves performance in terms of data rates over static allocation mechanisms by an additional 10%, and that allowing a slightly higher outage in not complying with the guaranteed data rate further increases the user's throughput by at least 20%.

While one of the main features of 5G networks is provisioning very high rates with low (or no) variability to cellular users, it has been shown that this turns out to be very ineffective for operators because it leads to an abundance of unused network resources. Yet, reallocating the unused resources to the same users, after providing them with the same constant rate, increases back the variability in data rates. A more efficient way would be to provide different low-variability data rates to the users depending on their channel conditions while trying to bring the wasted resources to the lowest possible extent. To that end, in this paper, two approaches are considered; one with reserved resources for every user and the other where the amount of resources is decided on the fly, depending on their current channel conditions. Then, for each approach, we look at different allocation policies and derive the corresponding maximum achievable constant rate for every user jointly with the level of resource utilization, showing which policy is more beneficial. Further, the performance is evaluated on a real 5G trace using both extensive simulations and real measurements conducted on OpenAirInterface. Results show that no-resource reservation policies increase the utilization of resources and data rates at the expense of increased rate variability across all the users. Moreover, all the policies proposed in this paper outperform state-of-the-art approaches by at least 2×, bringing the waste of resources down to 15%.

In our digitized society, emerging applications require highly-performing and flexible networks that can adapt to satisfy varying connectivity needs. P4 as a domain-specific programming language for data plane pipelines introduces the required flexibility through easy-to-use programmability. However, the performance of P4-capable devices is still an open question that has not yet been completely addressed. Understanding whether a P4-enabled device can meet the performance requirements for a specific network function pipeline is key for planning as well as for the selection of the proper deployment scenarios in a network. To bridge this gap, we propose a simple analytical model that can predict the performance of network functions written in P4 for a given device. The programmable data plane of P4 devices is modeled as a forward queueing system with a variable service rate that depends on the complexity of the configured data path program. On top of the data plane model, the controller’s interaction is modeled as a feedback queueing system. In terms of the analysis, we first assume exponentially distributed service times in the data plane and control plane. In a second step, we extend the analysis to generally distributed service times using approximations. In order to cover a wide rang of possible behavior of the control plane, three types of distributions with different coefficients of variation are inspected: Erlang, exponential, and hyperexponential. We evaluate the accuracy of our model for different scenarios and show that the discrepancy between actual results and our analytical predictions does not exceed 8.7%. We also validate the model with a commercial P4 hardware switch.

5G networks have emerged as the only viable solution to render a satisfying level of performance to different types of services, each of them with very stringent traffic requirements. One of those services are Ultra-Reliable Low-Latency Communications (URLLC). A use case where these services are especially sensitive are vehicular networks. Therefore, in order to satisfy their traffic requirements, adequate resource allocation schemes should be devised. However, the time-varying nature of the channel conditions in wireless networks renders this process challenging. In this paper, we consider the problem of jointly allocating Radio Access Network (RAN) resources and computing resources (to process the data from vehicles) such that all the traffic requirements of individual users are met and the utility is maximized for different types of fairness. We formulate an optimization problem for the general case of$\alpha$-fairness, explore its characteristics, and consider in more detail the opposite sides of fairness; the case of no fairness provided$(\alpha=0)$and the max-min fair allocation$(\alpha\rightarrow\infty)$. For each of these problems, we propose polynomial-time assignment heuristics. Using data from real traces, we show that the performance achieved with our approaches is not more than 1% away from the optimum.

The introduction of Software-Defined Radio Access Networks in 5G, whose main feature is the possibility of decoupling the control plane from the data plane, and associating the former with a controller away from base stations, represents a paradigm shift in the way the network resources are allocated. This property provides an increased flexibility in cellular network operation, yielding significant improvements compared to the pre-5G resource allocation era. However, the full extent to which this amelioration ranges is not yet clear for different metrics of interest and objectives. One such objective is to allocate resources so that proportional fairness is achieved. Therefore, in this paper, we consider analytically the problem of proportionally fair allocation in SD-RAN environments, by deriving the policy which accomplishes that. We do this for two scenarios. In the first, the goal is to provide proportional fairness across all the users in the network, whereas in the second, the objective is to provide proportionally fair allocation in terms of the throughput of all BSs. We evaluate the performance with input parameters from a real trace. Results show that the introduction of SD-RAN increases the value of the objective by up to an order of magnitude compared to the scenario with no SD-RAN.

Software-Defined Radio Access Networks (SD-RANs), introduced in 5G, represent a paradigm shift in the process of cellular network resource allocation. The decoupling of the control from the data plane, and associating the former with a controller away from Base Stations (BSs), has enabled an increased flexibility in allocating network resources which would lead to performance improvements. However, so far, it is not yet clear to what extent this amelioration ranges in terms of the maximum throughput that can be achieved. Therefore, in this paper, we consider analytically the problem of maximizing the overall network throughput in an SD-RAN environment, by deriving the policy which accomplishes that along with the total throughput, of interest to cellular operators. We assess the performance with real user traces. Results show that the introduction of SD-RAN improves performance by at least 20%.

Mobility management in 5G, especially at higher frequencies, is challenging because the signal quality fluctuates significantly due to blockages of Line of Sight (LoS), shadowing and user mobility. As a result, users experience frequent handovers, which reduce the network capacity. In order to perform smooth network operation, the decisions when to handover and to which Base Station (BS) a user is to be assigned should be considered jointly. Another important goal is to strive for fairness in data rates among the users. To this end, in this paper, we formulate an optimization problem whose solution provides proportional fairness and reduces the handover rate significantly. To solve the problem, we propose a Deep Reinforcement Learning (DRL) algorithm, specifically a Deep Q Network (DQN), which turns out to find a near-optimal user-to-BS assignment. We compare our approach with other state-of-the-art baselines and show that it outperforms them considerably in terms of fairness, handover, ping-pong and radio link failure rates while being within 96% of the optimal solution. Our DQN algorithm also reduces the handover rate by 86% and avoids ping-pong handovers.

The rising number of connected Internet of Things (IoT) devices in 5G networks and the standardization of the 3GPP reduced capability (RedCap) devices, turn the IoT energy efficiency into a topic of paramount importance for 5G. The design goals and use cases of RedCap devices highlight the need for long device battery life due to the infeasibility of replacing batteries. With the focus emerging on sustainable networks, battery lifetime prediction becomes essential. Therefore, in this paper, we propose and evaluate a Markov Chain based energy consumption model suitable for IoT devices in 5G networks, especially RedCap devices. We design a realistic model consistent with the procedures described in 3GPP standardization, mainly focused on the uplink transmission procedures. The proposed model is validated through extensive analysis with varying interarrival times (IAT) of the uplink traffic. For short IAT, the analytical results show a decrease of 33% in energy consumption and 89% in transmission latency. This demonstrates that our model can be applied to evaluate battery life for a broad range of IoT devices.

The possibility of decoupling the operation of control plane from data plane in RANs, which became possible with the introduction of Software-Defined Networks in 5G, brought a paradigm shift in cellular network operation. The key element that enables this is a centralized controller, located away from base stations. This yields increased flexibility in the functioning of cellular networks, resulting in considerable enhancements compared to classical pre-5G resource allocation approaches. However, so far the range these improvements span is known only in terms of throughput. The advantages in terms of other metrics and objectives, like delay fairness, are not yet known. Therefore, in this paper, we derive analytically the resource allocation policies that lead to different delay fairness definitions among the entities in an SD-RAN-enabled network and show the advantages compared to the classical pre-5G approaches. We do this for different scenarios. First, we consider the minimum potential delay fairness in the network. Then, we consider the min-max delay fairness among base stations, and also the min-max delay fairness among users. We evaluate performance extensively with input data from a dataset. The results indicate that the introduction of SD-RAN improves the objective value up to 6× compared to policies without SD-RAN.

One of the three types of services supported by 5G networks are Ultra-Reliable Low-Latency Communications, which are characterized by the stringent requirement to deliver packets within a very short time with a high reliability. Besides being successfully transmitted/received, these data need to be processed as well. To satisfy these strict requirements, one needs to determine both the required data rate and the processing rate, given the channel conditions and traffic intensity of the service. Moreover, with constraints on both the Radio Access Network and edge computing resources as well as with the competition between an ever-increasing number of users in cellular networks, a very important question which arises is that of admission control. This guarantees users will not suffer from deteriorating performance. In this paper, using analytical modeling, we derive admission control policies for both homogeneous and heterogeneous types of users, taking into account the delay incurred by the RAN part of the network and that caused by the finite computing capability at the edge. We validate theoretical outcomes and provide additional insights on a 5G dataset. Results show that the number of admitted users depends on the worst channel conditions, the deadline by which the data must be processed and the available resources. There is an almost linear increase in the number of admitted users with the decrease in latency.

Competitive demand for network resources has only increased during the emergence of 5G next generation cellular technology. As video streaming accounts for an overwhelming percentage of this demand, the importance of considering the often-neglected Quality of Experience (QoE) metric is essential to ensure network resources are allocated in the most effective manner. Generalized network throughput metrics are insufficient in capturing the full human experience as increased data rates do not necessarily translate to improvements in user utility. Our study compares the efficacy of existing network allocation algorithms and proposes new approaches to 5G network resource allocation schemes using a more inclusive snapshot of user demand. We provide recommendations on which approach provides the highest QoE performance and suggestions for future network-side improvements. We further propose a QoE-driven network resource allocation (QENA) algorithm that shows a 20% improvement in overall average QoE across a large set of heterogeneous users.

Mobility management in 5G is challenging, and at higher frequencies, a larger number of cells is needed to provide similar coverage to that in 4G. Consequently, Base Stations (BSs) are placed much more densely and users experience frequent handovers, reducing network capacity. Advanced handover techniques are needed in 5G to perform smooth network operation. In this paper, we formulate an optimization problem, whose goal is to strive for fairness in data rates among users and to reduce handovers. To accomplish that, we consider jointly the decisions when to handover and to which BS a user is to be assigned. This is an integer nonlinear program, and by relaxing it, we obtain an upper bound. Further, due to its NP-hardness, we propose a centralized and a multi-agent Deep Q Network (DQN)-based algorithm, which both find near-optimal user-to-BS assignments. We evaluate our Reinforcement Learning-based solutions for networks of different sizes and users with different velocities. We compare our approaches with baselines and show that they outperform them considerably in terms of fairness and radio link failures while being within 95% of the optimum. Our DQN algorithms also reduce the handover rate by up to 93% and avoid ping-pong handovers almost completely.

Crowdsourced image processing has the potential to vastly impact response timeliness in various emergency situations. Because images can provide extremely important information regarding an event of interest (hits), sending the right images to an analyzer as soon as possible is of crucial importance. In this paper, we consider the problem of optimally assigning resources, both local (CPUs in phones) and remote (network-based GPUs) to mobile devices for processing images, ultimately sending those of interest to a centralized entity while also accounting for the energy consumption at the distributed nodes. To that end, we use the dual-path Network Utility Maximization (NUM) framework, coupled with a hit-ratio estimator and energy costs, to enable a distributed implementation of the system. We include analysis of different hit-ratio estimators using realistic trace data, first considering immediate and then delayed feedback. We address accuracy concerns when estimating the likelihood of future imagehitsand provide a window-based heuristic for scenarios when hit-ratio feedback is severely delayed. Our TCP-inspired window-method predicts both imagehitlikelihood and current wireless network congestion with great effectiveness. Results are validated using both synthetic simulations and real-life traces.

Software-Defined Radio Access Networks (SD-RANs) foster the concepts of programmability and flexibility, which are vital for next generation cellular networks. However, SD-RANs render network management and orchestration very challenging. Indeed, related works indicate that when thousands of connected devices are spread across the underlying network, SD-RAN approaches with a single controller become deficient and exhibit undesired behavior. Despite this, state-of-the-art research papers lack concrete solutions and evaluations with respect to throughput predictability, where the latter is jeopardized by irregularities in the SD-RAN control plane, specifically in realistic testbeds. In order to overcome the aforementioned issues, in this work, we presentDelphi: a novel platform that provides both analytical and experimental methods to achieve our goal, which is computing the maximum achievable throughput in SD-RAN environments. Analyzing the results provided byDelphi, we can capture the impact of the SD-RAN control plane on throughput. Moreover, we can design important guidelines as to which policy to choose given objectives such as throughput maximization or robustness. Providing a platform for SD-RAN evaluations based on open-source components,Delphienables new avenues for research in the mobile network community. Focusing on FlexRAN SD-RAN controller for our initial results, overall, our findings show that when the number of Base Stations (BSs) and User Equipment (UEs) in the network increases beyond 5000, due to non-timely received control packets for the maximum Channel Quality Indicator (maxCQI) policy the overall throughput decreases by more than 20%.

As mobile devices become more prevalent in everyday life and the amount of recorded and stored videos increases, efficient techniques for searching video content become more important. When a user sends a query searching for a specific action in a large amount of data, the goal is to respond to the query accurately and fast. In this paper, we address the problem of responding to queries which search for specific actions in mobile devices in a timely manner by utilizing both visual and audio processing approaches. We build a system, called VidQ, which consists of several stages, and that uses various Convolutional Neural Networks (CNNs) and Speech APIs to respond to such queries. As the state-of-the-art computer vision and speech algorithms are computationally intensive, we use servers with GPUs to assist mobile users in the process. After a query is issued, we identify the different stages of processing that will take place. Then, we identify the order of these stages. Finally, solving an optimization problem that captures the system behavior, we distribute the process among the available network resources to minimize the processing time. Results show that VidQ reduces the completion time by at least 50% compared to other approaches.

Massive Machine-Type Communications (mMTC) are one of the three types of services that should be supported by 5G networks. These are distinguished by the need to serve a large number of devices which are characterized by non-intensive traffic and low energy consumption. While the sporadic nature of the mMTC traffic does not pose an exertion to efficient network operation, multiplexing the traffic from a large number of these devices within the cell certainly does. Therefore, planning carefully the network resources for this traffic is of paramount importance. To do this, the statistics of the traffic pattern that arrives at the base station should be known. To this end, in this paper, we derive the distribution of the inter-arrival times of the traffic at the base station from a general number of mMTC users within the cell, assuming a generic distribution of the traffic pattern by individual users. We validate our results on traces. Results show that adding more mMTC users in the cell increases the variability of the traffic pattern at the base station almost linearly, which is not the case with increasing the traffic generation rates.

Edge computing has become a very popular service that enables mobile devices to run complex tasks with the help of network-based computing resources. However, edge clouds are often resource-constrained, which makes resource allocation a challenging issue. We focus on a distributed resource allocation method in which servers operate independently and do not communicate with each other, but interact with clients (tasks) to make allocation decisions. This provides robustness and does not require service providers to share information about their configurations or workloads. We utilize a two-round bidding approach of assigning tasks to edge cloud servers. We consider a preemption-enabled system in which servers may stop a previous task in order to run a more useful one. We evaluate the performance of our system using realistic simulations and real-world trace data from a high-performance computing cluster. Results show that our approach is reasonably close to optimal assignment, while saving 50–70 % of the original computation time.

Software-Defined Networking (SDN) lays the foundation for the operation of future networking applications. The separation of the control plane from the programmable data plane increases the flexibility in network operation. One of the most used languages for describing the packet behavior in the data plane is P4. It allows protocol and hardware independent programming. With the expanding deployment of P4 programmable devices, it is of utmost importance to understand their performance behavior and limitations in order to design a network and provide Quality of Service (QoS) guarantees. One of the most important performance metrics is the packet mean sojourn time in a P4 device. While previous works already modeled the sojourn time in P4 devices with controller feedback, those models were rather simplified and could not capture the system behavior for general cases, resulting in a potential highly inaccurate performance prediction. To bridge this gap, in this paper, we consider the system behavior of P4 devices for the general case, i.e., under general assumptions. To that end, we model the behavior with a queueing network with feedback. As it is impossible to provide closed-form solutions, we consider different approximations for the mean sojourn time. We validate our results against extensive realistic simulations, capturing different behaviors in the data and control planes. Results show that the most accurate approximation in almost all cases is the one in which the queues are decoupled and considered as independent despite the fact that there are dependencies. The level of discrepancy in the worst case does not exceed 18.2% for service times distributions with a coefficient of variation not greater than 1.

Next generation radio access networks (RANs) en-vision softwarization and programmability as the main tools to provide the quality of service (QoS) requirements of emerging applications. Consequently, software-defined radio access networks (SD-RANs) have gained increased traction as a technology to foster network management and alleviate orchestration. While there exist SD-RAN architecture concepts both with single and multiple SD-RAN controllers, currently developed prototypes only include a single controller. Such a design may be sufficient for a low number of managed devices, for instance below 50. When the number of devices increases beyond 300, the controller performance deteriorates. A distributed control plane provides a solution, but renders the management in the control plane complex and incurs additional overhead, for instance control handover. In this way, both single controller and distributed control plane approaches may have a negative impact on a user’s QoS. Yet, proper evaluations are missing and therefore the performance remains unclear. In order to investigate the effect of SD-RAN control plane on the user performance, in this work, we provide an extensive evaluation based on a 5G simulator, compliant with 3GPP standardization, as well as measurements with open-source SD-RAN controllers. Based on our simulator, we are able to demystify the user QoS depending on the control plane design choices. Our results demonstrate that having a distributed control plane with control handovers improves the user performance by at least 20% in terms of throughput, 5x regarding the packet loss ratio and 140% in terms of delay compared to a single controller approach. This confirms that the benefits of multiple controllers surpass the overhead caused by more complicated management.

In the next-generation wireless networks, energy efficiency (EE) is a fundamental requirement due to the limited battery power and the deployment of various devices in hardly accessible areas. While a plethora of approaches have been proposed to increase users’ EE, there are still many unresolved issues stemming mainly from the limited wireless resources. In this paper, we investigate the energy-efficient resource allocation, taking into account users’ radio resource control (RRC) state. We aim to achieve max-min fairness among users in an uplink orthogonal frequency-division multiple access (OFDMA) system while fulfilling data rate requirements and transmit power constraints. In particular, we avoid waste of the energy through unnecessary state transitions when no network resources are available. We study the impact of the RRC Resume procedure on users’ EE and propose allocating resources while users are in their current RRC Connected or RRC Inactive state. The solution is obtained from a constrained optimization problem, whose output is max-min fair and energy-efficient. To that end, we use generalized fractional programming and the Lagrangian dual decomposition approach to allocate the radio resources and transmission power iteratively. Using extensive realistic simulations with input parameters from measurement data, we compare the results of our approach against benchmark models and show the performance improvements RRC state awareness brings. Specifically, using our approach, the users’ EE increases by at least 10% on average.

Consistent rate provisioning is one of the most prominent features envisioned for the next generation of cellular networks (5G), as a pivotal condition to an improved user experience, especially for services like live video streaming, online gaming, etc. However, prior research has shown that providing a consistent rate, while very beneficial to the QoS of mobile users, can result in a severe underutilization of the available resources, leading to a very inefficient operation of cellular networks. One of the ways of increasing resource allocation efficiency is by reallocating the unused resources to the same users. To this end, in this paper we quantify the benefits offered by different reallocation policies both for the mobile operator and users. We then determine, based on theoretical analysis, the optimal policies to follow for different optimization objectives. First, we focus on increasing the efficiency (total throughput) of the cellular network operator in a cell and then on providing proportional and max-min fairness to mobile users. The analysis captures the correlation of the user's channel quality in contiguous frames by using Markov chains. The outcomes of the analysis hold both for users that are always active in a given cell, and for users whose activity is intermittent. We also analyze the case with two classes of consistent users:premiumandregular. The theoretical analysis is validated by extensive synthetic simulations and simulations run on real-life traces. We also compare the performance of different reallocation policies with that of a benchmark and show that the optimal reallocation policy for a given objective improves the performance by at least 35 percent.

Multilevel security (MLS) systems control access to data by formalizing permissible and impermissible information flows between data sources and destinations (e.g., database servers and clients) fixed with distinct security labels. In computer networks, MLS systems have been used to prevent unauthorized data disclosure in shared-infrastructure settings where network hosts and devices may fall within different trust domains (e.g., in multi-tenant cloud networks or wireless mesh networks). However, current MLS systems assume static network behavior—thus preventing the network from being practically usable in the presence of dynamic network events that frequent unstable network environments, including sudden changes in traffic patterns, link failures, and topology changes as a result of device movement or intermittent device connectivity. In this paper, we introduceMLS-Enforcer, a software-defined networking (SDN) controller application that can efficiently deploy network-level MLS policies while retaining the ability to securely relabel network nodes under changing topology state and network traffic demands. We model network adaptivity as an integer linear programming problem that reflects a given security policy. We then introduce heuristic relabeling algorithms that achieve near-optimal performance and are more tractable and efficient for larger networks. We validateMLS-Enforceron several network topologies and traffic loads, demonstrating that it can relabel the network to route 90%+ of flows under normal conditions and quickly converge (on the order of seconds for the heuristic algorithms) under changing needs—from small network structure changes to catastrophic failures. This shows that formally secured networks can feasibly be deployed in diverse, changing, and unpredictable environments.

Crowdsourcing data collection from a network of mobile devices is useful in various applications. Mobile devices store a large amount of visual data that aid in different situations. Trained CNNs can be deployed on mobile devices to be used in searching for objects of interest. Querying for novel objects, for which models have not been trained, presents unique challenges. When novel objects are queried, new models must be trained and distributed to all edge devices, which can be cumbersome. In this paper we propose EDIR, an efficient method and a system that enables answering these queries while taking into account the bandwidth limitations in wireless networks, and the limited energy and computational power on mobile devices. Results show that EDIR reduces the amount of data transfer by 45%compared to other approaches while achieving a good F1 score.

Edge computing has become a very popular service that enables mobile devices to run complex tasks with the help of network-based computing resources. However, edge clouds are often resource-constrained, which makes resource allocation a challenging issue. We focus on a distributed resource allocation method in which servers operate independently and do not communicate with each other, but interact with clients (tasks) to make allocation decisions. This provides robustness and does not require service providers to share information about their configurations or workloads. We propose a two-round bidding approach of assigning tasks to edge cloud servers, while taking into account various processing requirements and server constraints. We consider cases in which all jobs have equal utility, cases where jobs have different utilities but users do not disclose these utilities to servers, and cases where users disclose the utility of their jobs to servers. We evaluate the performance using extensive realistic simulations. Results show that our approach is very close to an optimal assignment, with discrepancy not exceeding 5%.

The level of deployment of 5G networks is increasing every day, making this cellular technology become ubiquitous soon. Therefore, characterizing the channel quality and signal characteristics of 5G networks is of paramount importance as a first step in understanding the achievable performance of cellular users. Then, it can also serve for other important processes, such as resource planning and admission control. In this paper, we use the results of a publicly available measurement campaign of 5G users conducted by a third party and analyze various figures of merit. The analysis shows that the downlink and uplink rates for static and mobile users can be captured either by a lognormal or a Generalized Pareto distribution. Also, the time spent in the same cell by a mobile (driving) user can be captured to the best extent by a Generalized Pareto distribution. We also show some potential practical applications, among which is the prediction of the number of active users in the cell.

Ultra-Reliable Low-Latency Communications (URLLC) are one of the service types supported by 5G. These are characterized by a high reliability of delivering packets within a short deadline. To fulfill these stringent requirements, a special care must be taken to determine the required data rate of a user, given its traffic intensity and channel conditions. Furthermore, with the network resources being limited and the competition between the users in the cell, an important question that arises is that of admission control, so that the admitted users do not experience performance deterioration. In this paper, we provide the analysis that leads to an admission control policy. We do this for two types of users in terms of their traffic intensities and channel conditions: 1) homogeneous users, and 2) heterogeneous users. We validate our results on a trace. Results show that the number of admitted users depends on the traffic intensity and the worst channel conditions. An increase in traffic intensity by 3 times can decrease the number of admitted users by almost 35%.

Mobile devices collect a large amount of visual data that are useful for many applications. Searching for an object of interest over a network of mobile devices can aid human analysts in a variety of situations. However, processing the information on these devices is a challenge owing to the high computational complexity of the state-of-the-art computer vision algorithms that primarily rely on Convolutional Neural Networks (CNNs). Thus, this paper builds PicSys, a system that enables answering visual search queries on a mobile network. The objective of the system is to minimize the maximum completion time over all devices while taking into account the energy consumption of mobile devices as well. First, PicSys carefully divides the computation into multiple filtering stages, such that only a small percentage of images need to run the entire CNN pipeline. Splitting such CNN computation into multiple stages requires understanding the intermediate CNN features and systematically trading off accuracy for the computation speed. Second, PicSys determines where to run each of the stages of the multi-stage pipeline to fully utilize the available resources. Finally, through extensive experimentation, system implementation, and simulation, we show that PicSys performance is close to optimal and significantly outperforms other standard algorithms.

Mobile edge computing provides the opportunity for wireless users to exploit the power of cloud computing without a large communication delay. To serve data-intensive applications (e.g., video analytics, machine learning tasks) from the edge, we need, in addition to computation resources, storage resources for storing server code and data as well as network bandwidth for receiving user-provided data. Moreover, due to time-varying demands, the code and data placement needs to be adjusted over time, which raises concerns of system stability and operation cost. In this paper, we address these issues by proposing a two-time-scale framework that jointly optimizes service (code and data) placement and request scheduling, while considering storage, communication, computation, and budget constraints. First, by analyzing the hardness of various cases, we completely characterize the complexity of our problem. Next, we develop a polynomial-time service placement algorithm by formulating our problem as a set function optimization, which attains a constant-factor approximation under certain conditions. Furthermore, we develop a polynomial-time request scheduling algorithm by computing the maximum flow in a carefully constructed auxiliary graph, which satisfies hard resource constraints and is provably optimal in the special case where requests have homogeneous resource demands. Extensive synthetic and trace-driven simulations show that the proposed algorithms achieve 90% of the optimal performance.

Crowdsourced image processing has the potential to vastly impact response timeliness in various emergency situations. Because images can provide extremely important information regarding an event of interest, sending the right images to an analyzer as soon as possible is of crucial importance. In this paper, we consider the problem of optimally assigning resources, both local (CPUs in phones) and remote (network-based GPUs) to mobile devices for processing images, ultimately sending those of interest to a centralized entity while also accounting for the energy consumption. To that end, we use the Network Utility Maximization (NUM) framework, coupled with a hit-ratio estimator and energy costs, to enable a distributed implementation of the system. Our results are validated using both synthetic simulations and real-life traces.

Providing a consistent data rate to users with a highly dynamic activity in a cellular network is a very important feature of 5G. This will lead to increased user satisfaction with services such as video-streaming, online gaming, etc. Knowing that resources are constrained, operators must limit the number of users in the cell, so that admitted users do not experience performance deterioration. In this paper, we consider the problem of admission control for consistent users in 5G cellular systems. We consider two types of users in terms of their rate distributions: 1) homogeneous users, and 2) heterogeneous users. For each user type, we perform the analysis for two scenarios: constant number of users, and random number of users. The analysis is followed by realistic simulations to show that our models can provide satisfactory results in realistic scenarios as well. We also show that the number of admitted users can be significantly increased (up to 3 ×) by allowing a slight deterioration in the QoS.

Consistent rate provisioning is one of the most prominent features envisioned in next generation of cellular networks (5G). However, it has been shown that providing a consistent data rate to users leads to severe underutilization of the available resources, making the cellular operator function very inefficiently. A possible way to increase the efficiency is by reallocating the unused resources to the same users. In this paper, we quantify the benefits offered by different reallocation policies both for mobile operator and the users. First, we focus on increasing the efficiency (total throughput) of the cellular network operator and then on providing fairness to mobile users. In both cases, we determine the optimal policy. The theoretical analysis is followed by extensive realistic simulations, where we compare the performance of different reallocation policies with that of a benchmark model (fair share of the resources with no consistency), and show that the right reallocation policy improves the performance significantly.

Mobile edge computing allows wireless users to exploit the power of cloud computing without the large communication delay. To serve data-intensive applications (e.g., augmented reality, video analytics) from the edge, we need, in addition to CPU cycles and memory for computation, storage resource for storing server data and network bandwidth for receiving user-provided data. Moreover, the data placement needs to be adapted over time to serve time-varying demands, while considering system stability and operation cost. We address this problem by proposing a two-time-scale framework that jointly optimizes service (data & code) placement and request scheduling, under storage, communication, computation, and budget constraints. We fully characterize the complexity of our problem by analyzing the hardness of various cases. By casting our problem as a set function optimization, we develop a polynomial-time algorithm that achieves a constant-factor approximation under certain conditions. Extensive synthetic and trace-driven simulations show that the proposed algorithm achieves 90% of the optimal performance.

Providing a consistent data rate to mobile users will be a very important feature of next generation systems, i.e., 5G, especially for services such as live video-streaming, online gaming, etc. This could lead to an increased user satisfaction with these services. In this paper, we perform the analysis to determine the maximum consistent data rate that can be offered to a (high paying) class of mobile users, both within a cell and within a region covered with multiple cells, given certain available resources. We do this for two cases: 1) when the number of active users in the class is constant, and 2) for a varying number of users being simultaneously present and active in the class. The analysis is performed under some independence assumptions, but we validate our results with extensive realistic simulations where the assumptions are relaxed. We show that providing consistent rate is rather expensive because a large percentage of the available resources remain unused most of the time. However, the unused resources can be shared (possibly equally) by the users in the group. In that case the consistent rate can be seen as a guaranteed minimum rate. The other option is to allocate the unused resources to a class of best effort users. We show that using any of these options will result in significant performance improvements.

We consider a scenario in which an operator installs (small cell) base stations on top of city buses to offer better quality of experience (QoE) to their passengers. In that case, providing a consistent backhaul rate (i.e., a constant rate at all times) to these base stations could help mitigate the effects of mobility on the QoE. Specifically, we perform the analysis to determine the maximum consistent backhaul rate that can be offered to a bus on a given route, given the resources allocated by the operator to backhauling, by taking advantage of the fact that different buses on that route will see different conditions at a given time. We also consider the case where we allow a small outage probability, i.e., that the consistent rate is not provided for a small proportion of time. We show that by allowing an outage probability of only 1% we can increase the achievable backhaul rate by 50%. We then show how to compute the Pareto frontier (rate region) of the achievable consistent backhaul rates when there are two bus routes. The analysis is performed under an independence assumption and hence we validate our results by simulations. Altogether, the cost of consistency is very high, but it can be partly mitigated by allocating the unused backhaul capacity to best effort services in real time.

An unprecedented increase in the mobile data traffic volume has been recently reported due to the extensive use of smartphones, tablets, and laptops. This is a major concern for mobile network operators, who are forced to often operate very close to their capacity limits. Recently, different solutions have been proposed to overcome this problem. The deployment of additional infrastructure, the use of more advanced technologies (LTE), or offloading some traffic through Femtocells and WiFi are some of the solutions. Out of these, WiFi presents some key advantages such as its already widespread deployment and low cost. While benefits to operators have already been documented, it is less clear how much and under what conditions the user gains as well. Additionally, the increasingly heterogeneous deployment of cellular networks (partial 4G coverage, small cells, etc.) further complicates the picture regarding both operatorand user-related performance of data off loading. To this end, in this paper we propose a queueing analytic model that can be used to understand the performance improvements achievable by Wi-Fi-based data off loading, as a function of Wi-Fi availability and performance, user mobility and traffic load, and the coverage ratio and respective rates of different cellular technologies available. We validate our theory against simulations for realistic scenarios and parameters, and provide some initial insights as to the offloading gains expected in practice.

Operators have recently resorted to Wi-Fi offloading to deal with increasing data demand and induced congestion. Researchers have further suggested the use of delayed offloading: if no Wi-Fi connection is available, (some) traffic can be delayed up to a given deadline or until WiFi becomes available. Nevertheless, there is no clear consensus as to the benefits of delayed offloading, with a couple of recent experimental studies largely diverging in their conclusions, nor is it clear how these benefits depend on network characteristics (e.g., Wi-Fi availability), user traffic load, and so on. In this paper, we propose a queueing analytic model for delayed offloading, and derive the mean delay, offloading efficiency, and other metrics of interest, as a function of the user's patience, and key network parameters for two different service disciplines (First Come First Served and Processor Sharing). We validate the accuracy of our results using a range of realistic scenarios and real data traces. Finally, we use these expressions to show how the user could optimally choose deadlines by solving the variations of a constrained optimization problem, in order to maximize her own benefits.

Operators have recently resorted to WiFi offloading to deal with increasing data demand and induced congestion. Researchers have further suggested the use of “delayed offloading”: if no WiFi connection is available, (some) traffic can be delayed up to a given deadline, or until WiFi becomes available. Nevertheless, there is no clear consensus as to the benefits of delayed offloading, with a couple of recent experimental studies largely diverging in their conclusions. Nor is it clear how these benefits depend on network characteristics (e.g. WiFi availability), user traffic load, etc. In this paper, we propose a queueing analytic model for delayed offloading, and derive the mean delay, offloading efficiency, and other metrics of interest, as a function of the user's “patience”, and key network parameters. We validate the accuracy of our results using a range of realistic scenarios, and use these expressions to show how to optimally choose deadlines.

An unprecedented increase in the mobile data traffic volume has been recently reported due to the extensive use of smartphones, tablets and laptops. Moreover, predictions say that this increase is going to be yet more pronounced in the next 3-4 years. This is a major concern for mobile network operators, who are forced to often operate very close to (or even beyond) their capacity limits. Recently, different solutions have been proposed to overcome this problem. The deployment of additional infrastructure, the use of more advanced technologies (LTE), or offloading some traffic through Femtocells and WiFi are some of the solutions. Out of these, WiFi presents some key advantages such as its already widespread deployment and low cost. While the benefits to operators have already been documented, with considerable amounts of traffic already switched over to WiFi, it is less clear how much and under what conditions the user gains as well. To this end, in this paper we propose a queueing analytic model that can be used to understand the performance improvements achievable by WiFi-based data offloading, as a function of WiFi availability and performance, and user mobility and traffic load. We validate our theory against simulations for realistic data and scenarios, and provide some initial insights as to the offloading gains expected in practice.

Cognitive Networks have been proposed to opportunistically discover and exploit (temporarily) unused licensed spectrum bands. With the exception of TV white spaces, secondary users (SUs) can access the medium only intermittently, due to deferring to primary user (PU) transmissions and scanning for new channels. This raises the following questions: (i) what sort of delays can an SU expect on a channel given the PU utilization of this channel? (ii) how do specific characteristics of the PU activity patterns (e.g. burstiness) further affect performance? These questions are of key importance for the design of efficient algorithms for scheduling, spectrum handoff, etc. In this paper, we propose a queueing analytical model to answer them. We model the PU activity pattern as an ON-OFF alternating renewal process with generic ON and OFF durations, and derive a closed form expression for packet delays by solving a variant of the M/G/1 queue. Contrary to the common belief that low utilization channels are good channels, we show that the expected SU delay on a channel, and thus the best channel to use, is a subtle interplay between the ON and OFF duration distributions of the primary users, and the SU traffic load. We validate our analysis against simulations for different PU activity profiles.

Cognitive Networks have been proposed to opportunistically discover and exploit (temporarily) unused licensed spectrum bands. For a number of applications, high throughput is the key figure of merit, while the application is still elastic enough to be supported at different rates. To this end, the cognitive node will try to discover and pool together a number of (at the time available) primary channels to provide a given target throughput. When a single radio is used for both transmission and channel scanning, an interesting tradeoff arises: when one or more channels of the currently available ones are lost (e.g. primary user returns), should the node start scanning immediately or continue transmitting over the remaining channels. Using renewal-reward theory, we show that if the goal is to maximize the average (long-term) throughput, the answer to this question depends on the statistics of the channel availability periods. Specifically, for relatively homogeneous channels, we show that it is optimal to start scanning immediately, while for heterogeneous channels, it is often better to defer scanning, even if multiple channels are lost. Simulations for a range of different channel characteristics validate our analytical findings and suggest that triggering the scanning function at the right times, can improve performance considerably.