Services rely more and more on cloud platforms to offer their products to end users. This implies that being able to estimate the latency required to reach those cloud platforms is of growing importance. To shed light on this crucial aspect, we perform a three-month measurement campaign involving traceroute measurements every 30 minutes over 256 pairs of source-destination probes, where the vantage points are located in different Cloud Service Providers (CSPs) and the destination probes belong to one of the main Infrastructure Operators (IOs) of Spain. We provide interesting insights obtained from analyzing the data resulting from this campaign. Among them, we observe that, as expected, distance is the unavoidable factor impacting cloud latency. Yet, other results are less anticipated, such as the great stability of the network, or the lack of performance difference when comparing standard and premium network service tiers. We also analyze the potential of forecasting the cloud latency both for future samples but also for unobserved connections.

This work studies a well-known shared-cache coded caching scenario where each cache can serve an arbitrary number of users. We analyze the case where there is some knowledge about such number of users (i.e., the topology) during the content placement phase. Under the assumption of regular placement and a cumulative cache size that can be optimized across the different caches, we derive the fundamental limits of performance by introducing a novel cache-size optimization and placement scheme and a novel information-theoretic converse. The converse employs new index coding techniques to bypass traditional uniformity requirements, thus finely capturing the heterogeneity of the problem, and it provides a new approach to handle asymmetric settings. The new fundamental limits reveal that heterogeneous topologies can in fact outperform their homogeneous counterparts where each cache is associated to an equal number of users. These results are extended to capture the scenario of topological uncertainty where the perceived/estimated topology does not match the true network topology. This scenario is further elevated to the stochastic setting where the user-to-cache association is random and unknown, and it is shown that the proposed scheme is robust to such noisy or inexact knowledge on the topology.

Zero-touch network management is one of the most ambitious yet strongly required paradigms for beyond 5G and 6G mobile communication systems. Achieving full automation requires a closed loop that combines (i) network status data collection and processing, (ii) predictive capabilities based on such data to anticipate upcoming needs, and (iii) effective decision making that best addresses such future needs through proper network control and orchestration. Recent seminal works have proposed approaches to jointly implement the last two phases above via a single deep learning model trained on past network status to directly optimize future decisions. This is achieved by designing custom loss functions that directly embed the management task objective. Experiments with real-world measurement data have demonstrated that this strategy leads to substantial performance gains across diverse network management tasks. In this paper, we go one step beyond the loss tailoring schemes above, and introduce a loss meta-learning paradigm that (i) reduces the need for human intervention at model design stage, (ii) eases explainability and transferability of trained deep learning models for network management, and (iii) outperforms custom losses across a range of controlled experiments and practical use cases.

We analyze 4G and 5G transport-layer sessions generated by a wide range of mobile services at over 282,000 base stations (BSs) of an operational mobile network, and carry out a statistical characterization of their demand rates, associated traffic volume and temporal duration. Based on the gained insights, we model the arrival process of sessions at heterogeneously loaded BSs, the distribution of the session-level load and its relationship with the session duration, using simple yet effective mathematical approaches. Our models are fine-tuned to a variety of services, and complement existing tools that mimic packet-level statistics or aggregated spatiotemporal traffic demands at mobile network BSs. They thus offer an original angle to mobile traffic data generation, and support a more credible performance evaluation of solutions for network planning and management. We assess the utility of the models in practical application use cases, demonstrating how they enable a more trustworthy evaluation of solutions for the orchestration of sliced and virtualized networks.

A variety of network management and orchestration (MANO) tasks take advantage of predictions to support anticipatory decisions. In many practical scenarios, such predictions entail two largely overlooked challenges: (i) the exact relationship between the predicted values (e.g., reserved resources) and the performance objective (e.g., quality of experience of end users) is often tangled and cannot be known a priori, and (ii) the objective is linked in many cases to multiple predictions that contribute to it in an intertwined way (e.g., resources to reserved are limited and must be shared among competing flows). We present AutoManager, a novel meta-learning model that can support complex MANO tasks by addressing these two challenges. Our solution learns how multiple intertwined predictions affect a common performance goal, and steers them so as to attain the correct operation point under a-priori unknown loss functions. We demonstrate AutoManager in practical, complex use cases based on real-world traffic measurements; our experiments show that the model produces forecasts that are accurate and tailored to the MANO task in a fully automated way.

Cloud-native mobile networks pave the road for Network Slicing as a Service (NSaaS), where slice overbooking is a promising management strategy to maximize the revenues from admitted slices by exploiting the fact they are unlikely to fully utilize their reserved resources concurrently. While seminal works have shown the potential of overbooking for NSaaS in simplistic cases, its realization is challenging in practical scenarios with realistic slice demands, where its actual performance remains to be tested. In this paper, we propose kaNSaaS, a complete solution for NSaaS management with slice overbooking that combines deep learning and classical optimization to jointly solve the key tasks of admission control and resource allocation. Experiments with large-scale measurement data of actual tenant demands show that kaNSaaS increases the network operator profits by 300% with respect to NSaaS management strategies that do not employ overbooking, while outperforming by more than 20% state-of-the-art overbooking-based approaches.

The development of novel use cases in beyond-5G and 6G networks will rely, among other aspects, on the availability of computing resources at the edge, therefore enabling the realization of applications that are both computationally demanding and latency constrained, such as Mobile Augmented Reality (MAR). Indeed, due to end devices' intrinsic constraints on computation capabilities and battery, newer MAR applications require offloading their most demanding tasks. However, the constrained nature of edge resources implies that these tasks should be carefully allocated at the edge network in order to guarantee satisfactory Quality of Experience to end-users. In this context, we analyze the edge operator's resource allocation to support the energy-aware offloading of MAR tasks at the edge of the cellular network with the goal of not only maximizing service acceptance (i.e., revenue), but also optimizing the operator's business utility, which depends on its carbon footprint and the profit of operating the service. We leverage Deep Reinforcement Learning to propose an efficient model to operate the edge resource allocation that can adapt to different utilities.

The recent introduction of vector coded caching has revealed that multi-rank transmissions in the presence of receiver-side cache content can dramatically ameliorate the file-size bottleneck of coded caching and substantially boost performance in error-free wire-like channels. In this work, we employ large-matrix analysis to explore the effect of vector coded caching in realistic wireless multi-antenna downlink systems. For a given downlink MISO system already optimized to exploit both multiplexing and beamforming gains, and for a fixed set of antenna and SNR resources, our analysis answers a simple question: What is the multiplicative throughput boost obtained from introducing reasonably-sized receiver-side caches that can pre-store information content? The derived closed-form expressions capture various linear precoders, and a variety of practical considerations such as power dissemination across signals, realistic SNR values, as well as feedback costs. The schemes are very simple (we simply collapse precoding vectors into a single vector), and the recorded gains are notable. For example, for 32 transmit antennas, a received SNR of 20 dB, a coherence bandwidth of 300 kHz, a coherence period of 40 ms, and under realistic file-size and cache-size constraints, vector coded caching is here shown to offer a multiplicative throughput boost of about 310% with ZF/RZF precoding and a 430% boost in the performance of already optimized MF-based (cacheless) systems. Interestingly, vector coded caching also accelerates channel hardening to the benefit of feedback acquisition, often surpassing 540% gains over traditional hardening-constrained cacheless downlink systems.

We investigate the performance of coded caching in land mobile-satellite (LMS) systems, where a satellite station with full access to a content library serves K cache-aided land users. The promising gains that coded caching provides for the error-free shared-link Broadcast Channel are known to suffer in some low signal-to-noise ratio (SNR) scenarios. For this reason, we analyze to what extent the coded caching gains are preserved in LMS systems, which are governed by low-to-moderate SNR due to the long propagation distances. We model the satellite-terrestrial channels through the widely adopted Shadowed-Rician fading, and we show that the coded caching gains are partially preserved even in the low-SNR limit due to the existence of line-of-sight (LOS) components, which allows us to double the goodput of LMS communication at low SNR. These results illustrate the potential of coded caching in LMS systems and motivate further research to design adapted and practical coded caching schemes for LMS systems.

Multi-antenna cache-aided wireless networks were thought to suffer from a severe feedback bottleneck, since achieving the maximal Degrees-of-Freedom (DoF) performance required feedback from all served users for the known transmission schemes. These feedback costs match the caching gains and thus scale with the number of users. In the context of the$L$-antenna Multiple-Input Single Output broadcast channel with$K$receivers, each having normalized cache size$\gamma $, we pair a fundamentally novel algorithm together with a new information-theoretic converse and identify the optimal tradeoff between feedback costs and DoF performance, by showing that having channel state information from only$C< L$served users implies an optimal one-shot linear DoF of$C+K\gamma $. As a side consequence of this, we also now understand that the well known DoF performance$L+K\gamma $is in fact exactly optimal. In practice, the above means that we are able to disentangle caching gains from feedback costs, thus achieving unbounded caching gains at the mere feedback cost of the multiplexing gain. This further solidifies the role of caching in boosting multi-antenna systems; caching now can provide unbounded DoF gains over multi-antenna downlink systems, at no additional feedback costs. The above results are extended to also include the corresponding multiple transmitter scenario with caches at both ends.

In this paper, we analyze the high-SNR regime of the$M\times K$Network MISO channel in which each transmitter has access to a different channel estimate, possibly with different precision. It has been recently shown that, for some regimes, this setting attains the same Degrees-of-Freedom as the ideal centralized setting with perfect Channel State Information (CSI) sharing, in which all the transmitters are endowed with the best estimate available at any transmitter. This result is restricted by the limitations of the Degrees-of-Freedom metric, as it only provides information about the slope of growth of the capacity as a function of the SNR, without any insight about the possible performance at a given SNR. In order to overcome this limitation, we analyze the affine approximation of the rate on the high-SNR regime for this decentralized Network MISO setting for the antenna configurations in which it achieves the Degrees-of-Freedom of the centralized setting. We show that, for a regime of antenna configurations, it is possible to asymptotically attain the same achievable rate as in the ideal centralized scenario. Consequently, it is possible to achieve the beamforming gain of the ideal perfect-CSI-sharing setting even if only a subset of transmitters is endowed with precise CSI, which can be exploited in scenarios such as distributed massive MIMO where the number of transmit antennas is much bigger than the number of served users. This outcome is a consequence of the synergistic compromise between CSI precision at the transmitters and consistency between the locally-computed precoders, which is an inherent trade-off of decentralized settings that does not exist in the centralized CSI configuration. We propose a precoding scheme achieving the previous result, which is built on an uneven structure in which some transmitters reduce the precision of their own precoding vector for the sake of using transmission parameters that can be more easily predicted by the other transmitters.

We address the worst-user bottleneck of wireless coded caching, which is known to severely diminish cache-aided multicasting gains due to the fundamental worst-channel limitation of multicasting. We consider the quasi-static Rayleigh fading Broadcast Channel, for which we first show that the effective coded caching gain of the standard XOR-based coded-caching scheme completely vanishes in the low signal-to-noise ratio (SNR) regime. Then, we reveal that this collapse is not intrinsic to coded caching. We do so by presenting a novel scheme that can fully recover the coded caching gains by capitalizing on one aspect that has remained unexploited to date: the shared side information brought about by the effectively unavoidable file-size constraint. As a consequence, the worst-user effect is dramatically ameliorated, as it is substituted by a much more subtle worst-group-of-users effect, where the suggested grouping is fixed, and it is decided before the channel or the demands are known. Furthermore, the theoretical gains are completely recovered as the number of users increases, and this is done without any user selection technique. We analyze the rate performance of the proposed scheme and derive approximations which prove to be very precise. Importantly, this novel approach can be translated to other coded caching schemes and scenarios, including decentralized scenarios.

We investigate the use of coded caching in the single-cell downlink scenario where the receiving users are randomly located inside the cell. We first show that, as a result of having users that experience very different path-loss, the real gain of the original coded caching scheme is severely reduced. We then prove that the use of shared caches - which, we stress, is a compulsory feature brought about by the subpacketization constraint in nearly every practical setting - introduces a spatial-averaging effect that allows us to recover most of the subpacketization-constrained gains that coded caching would have yielded in the error-free identical-link setting. For the ergodic-fading scenario with different pathloss, we derive tight approximations of the average (over the users) rate and of the coded caching gain by means of a basic high-SNR approximation on the point-to-point capacity. These derived expressions prove very accurate even for low SNR. We also provide a result based on the regime of large number of users which is nonetheless also valid for settings with few users. These results are extensively validated using Monte-Carlo simulations that adhere to 3GPP recommendations on system parameters for urban micro or macro cells.

We address the problem of resource allocation (RA) for spectrum underlay in a cognitive radio (CR) communication system with multiple secondary operators sharing resource with an incumbent primary operator. The multiple secondary operator RA problem is well known to be especially challenging because of the inter-operator coupling constraints arising in the optimization problem, which render impractical inter-operator information exchange necessary. In this paper, we consider a satellite setting for multi-operator CR. In the CR maturation regime, i.e., the period in which the secondary subscriber density is growing yet remains much below that of incumbent users, we show that in fact the inter-operator mutual constraints can be neglected, thus making distributed (across secondary operators) optimization possible. Furthermore, we establish analytically that the mutual constraints asymptotically vanish with the primary user density.

In this work, we study the impact of imperfect sharing of the Channel State Information (CSI) available at the transmitters on a Network MIMO setting in which a set of M transmit antennas, possibly not co-located, jointly serve two multi-antenna users endowed with N1and N2antennas, respectively. We consider the case where only a subset of k transmit antennas have access to perfect CSI, whereas the other M - k transmit antennas have only access to finite precision CSI. The analysis of this configuration aims to answer the question of how much an extra informed antenna can help. We model this scenario as a Decentralized MIMO Broadcast Channel (BC) and characterize the Degrees-of-Freedom (DoF) region, showing that only k = max(N1, N2) antennas with perfect CSI are needed to achieve the DoF of the conventional BC with ubiquitous perfect CSI.

One of the famous problems in communications was the so-called "PN" problem in the Broadcast Channel, which refers to the setting where a fixed set of users provide perfect Channel State Information (CSI) to a multi-antenna transmitter, whereas the remaining users only provide finite precision CSI or no CSI. The Degrees-of-Freedom (DoF) of that setting were recently derived by means of the Aligned Image Set approach. In this work, we resolve the cache-aided variant of this problem (i.e., the "PN" setting with side information) in the regime where the number of users providing perfect CSI is smaller than or equal to the number of transmit antennas. In particular, we derive the optimal rate-memory trade-off under the assumption of uncoded placement, and characterize the same trade-off within a factor of 2.01 for general placement. The result proves that the "PN" impact remains similar even in the presence of side information, but also that the optimal trade-off is not achievable through serving independently the two sets of users.

In this work, we address the worst-user bottleneck of coded caching, which is known to diminish any caching gains due to the fundamental requirement that the multicast transmission rate should be limited by that of the worst channel among the served users. We consider the quasi-static Rayleigh fading Broadcast Channel, for which we first show that the coded caching gain of the XOR-based standard coded-caching scheme completely vanishes in the low-SNR regime. Yet, we show that this collapse is not intrinsic to coded caching by presenting a novel scheme that can completely recover the caching gains. The scheme exploits an aspect that has remained unexploited: the shared side information brought about by the file size constraint. The worst-user effect is dramatically ameliorated because it is replaced by the worst-group-of-users effect, where the users within a group have the same side information and the grouping is decided before the channel or the demands are known.

We study the Degrees-of-Freedom (DoF) in a wireless setting in which K Transmitters (TXs) aim at jointly serving K users. The performance is studied when the TXs are faced with a distributed Channel State Information (CSI) configuration in which each TX has access to its own multi-user imperfect channel estimate based on which it designs its transmit coefficients. The channel estimates are not only imperfectly acquired but they are also imperfectly shared between the TXs. Our first contribution consists of computing a genie-aided upper bound for the DoF of that setting. Our main contribution is then to develop a new robust transmission scheme that leverages the different qualities of CSI available at the TXs to improve the achieved DoF. We show the surprising result that there is a CSI regime, coined the Weak-CSIT regime, in which the genie-aided upper bound is achieved by the proposed transmission scheme. Interestingly, the optimal DoF in the Weak-CSIT regime only depends on the CSI quality at the best informed TX and not on the CSI quality at all other TXs.

In this paper1, we analyze a Network MIMO channel with 2 Transmitters (TXs) jointly serving 2 users, where each TX has a different multi-user Channel State Information (CSI), potentially with a different accuracy. Recently it was shown the surprising result that this decentralized setting can attain the same Degrees-of-Freedom (DoF) as its genie-aided centralized counterpart in which both TXs share the best-quality CSI. However, the DoF derivation alone does not characterize the actual rate and the question was left open as to how big the rate gap between the centralized and the decentralized settings was going to be. In this paper, we considerably strengthen the previous intriguing DoF result by showing that it is possible to achieve asymptotically the same sum rate as that attained by Zero-Forcing (ZF) precoding in a centralized setting endowed with the best-quality CSI. This result involves a novel precoding scheme which is tailored to the decentralized case. The key intuition behind this scheme lies in the striking of an asymptotically optimal compromise between i) realizing high enough precision ZF precoding while ii) maintaining consistent-enough precoding decisions across the non-communicating cooperating TXs.

This work1analyzes the Generalized Degrees-of-Freedom (GDoF) of the 2-User Multiple-Input Single-Output (MISO) Broadcast Channel (BC) in the so-called Distributed CSIT regime, with application to decentralized wireless networks. This regime differs from the classical limited CSIT one in that the CSIT is not just noisy but also imperfectly shared across the transmitters (TXs). Hence, each TX precodes data on the basis of local CSIT and statistical quality information at other TXs. We derive the GDoF result and obtain the surprising outcome that by specific accounting of the pathloss information, it becomes possible for the decentralized precoded network to reach the same performance as a genie-aided centralized network where the central node has obtained the estimates of both TXs. The key idea allowing this surprising robustness is to let the TXs have asymmetrical roles such that the most informed TX is able to balance the lower CSIT quality at the other TX.