Telecom Networks

Network slicing allows Mobile Network Operators (MNOs) to partition their physical infrastructure into multiple virtual logical networks, enabling the simultaneous servicing of applications with diverse Quality of Service characteristics. In the contenxt of H2020 5G-VINNI project, we introduced and evaluated economic models and policies for the provisioning, across multiple MNOs, of network slice services to Application Providers. We introduced a Network-Slice-as-a-Service model that maps the service offered by a network slice to requirements on virtualized resources. The placement of virtualized resources over the physical infrastructure of MNOs is determined by an embedding problem, formulated as a Mixed Integer Program. We investigate the embedding under: (i) centralized approaches, where a central Broker determines the embedding for all network slice requests, and (ii) peer-topeer approaches, where each MNO determines the embedding for the sub-set of network slice requests coming from its own customers. We introduced policies for cooperative modes, with the objective of total profit maximization, and for “coopetitive” (cooperative competition) modes, where MNOs aim to maximize their individual profits. The numerical results revealed that MNOs can maximize their aggregate and individual profits under any approach or mode, if they comply with the proposed policies.

The availability of Experimentation as a Service (EaaS) has been crucial in 5G research/market implementation. In the contenxt of H2020 5G-VINNI project, we introduced and evaluated novel business models for 5G EaaS, drawing on insights from the pre-commercial 5G-VINNI experimentation platform. Relying on the concept of value network, we developed a value network for 5G services with relevant business roles and relationships. On the basis of this, we presented the different actor roles involved in the provisioning of 5G EaaS, and we investigated alternative business models the actors can combine their offerings. Our analysis has shown that when a 5G vertical market (e.g., e-health, automotive, etc.) is immature, an aggressive business model is more attractive for Experimental Infrastructure Operators, while in the opposite case a conservative strategy is preferable. Finally, our results provide useful insights for the potential business models of a future commercial 5G/6G EaaS market.

The inherently multi-stakeholder value chain of 5G services calls for business and service coordination. In H2020 5GEx projext, we introduced and evaluated coordination models for the multi-provider service composition, namely the Fully Centralized, Distributed and per-Provider Centralized models, in the context of the 5GEx multi-provider orchestration framework. We performed a scalability assessment of the models in terms of the message overhead, also investigating the trade-off between service composition effciency and message overhead. Our sensitivity analysis on the different parameters of our evaluation framework revealed that hybrid models scale better, but also other models may achieve the same level of message overhead under certain conditions.

In the context of EU FP7 SmartenIT project, we introduced the Inter-Cloud Communication Traffic Management mechanism (ICC). ICC performs rate control over the ISP transit link(s) aiming to attain a target reduction of transit cost. In particular, ICC reduces Network Service Provider’s (NSP’s) transit charge by shaping a portion of the inter-domain traffic (e.g. delay-tolerant inter-cloud traffic), which has been marked as time-shiftable by the traffic source, i.e. by the business customer of the NSP such as a Cloud Service Provider or a Datacenter Operator. Furthermore, ICC reduces the transmission rate of marked traffic at peak 5-min billable intervals and increases it at off-peak, acting at even shorter timescales according to a novel rate-adaptation algorithm. Numerical evaluations of ICC by employing real traffic traces have revealed that ICC can significantly reduce the ISP transit charge and thus can be a promising and practically applicable solution for ISPs.

In the context of EU FP7 SmartenIT project , we specified the problem of the formation of an sustainable computational resource federation by Cloud Service Providers (CSPs) aiming to provide Quality of Service (QoS) to federation customers, expressed in terms of average total delay per job. Evaluating the economics of such a federation, we specified a model for the characterization of interactions among CSPs within a cloud federation. We modeled the energy consumption cost of each CSP as a function of its resource utilization factor, and the CSP's revenue by a delay-dependent pricing function according to which each CSP charges its clients. Finally, we formulated the problem of finding the federation policy that maximizes the total profit (revenue minus cost) of CSPs and we dealt with the incentives of individual CSPs. Numerical results demonstrated that interacting CSPs within a cloud federation can indeed maximize the total profit of the federation and also achieve a nearly optimal QoS.

One of the main intentions of the TRILOGY project was to develop new solutions in order to deliver effective and efficient control of resource sharing. An important outcome was the deployment of a new transport protocol, called Multipath TCP (MPTCP). MPTCP is a TCP variant that transmits data between two network nodes along multiple paths at the same time, in order to pool the capacity and reliability of multiple network paths and to benefit from multi-homing at endpoints – for example a mobile device with several wireless interfaces. An open source version of this protocol is available for the Linux operating system, which has been tested and ported to Android and Nokia devices. In addition, the ICT company Oracle is implementing MPTCP for its Solaris operating system. Updated versions of MPTCP have already been included in additional mobile devices.

 Please follow the links below for more information.

Linux Kernel MultiPath TCP project     Future Internet Award   Press [in greek]

In the context of EU FP7 SmartenIT project, we proposed an ISP-friendly mechanism for enhancing content delivery exploiting social information extracted from OSNs, e.g. social relationships, common interests and locality of content exchange. Constituent elements of SEConD are: i) a socially-aware proxy server inserted in a local geographic region (e.g. an AS) to orchestrate content distribution, ii) socially-aware messaging overlays employed to trigger video prefetching, iii) content-based P2P overlays employed to perform video streaming in each region, and iv) a two-level caching strategy both in the socially-aware proxy server and in the OSN user’s device whenever online. We also developed an evaluation framework to simulate the generation of content in the environment of an OSN, and to evaluate our mechanism and compare it with other approaches in the literature. Evaluation results showed that SEConD improves users’ Quality of Experience (QoE) and simultaneously, reduces traffic in potentially expensive inter-domain links, as well as the origin content server contribution.

In the context of EU FP7 SmoothIT project, we proposed and investigated the insertion of high-bandwidth ISP-owned Peers (IoPs) as an optimization approach for distributed content delivery so as to improve end-users’ performance and reduce inter-domain traffic. The motivation for this approach stems from the tit-for-tat mechanism applied by the choking algorithm of BitTorrent, i.e., the famous content delivery overlay network, and the fact that the IoP is equipped by the ISP with abundant network (and storage) resources. Then, due to the high upload capacity of the IoP, regular peers are found to establish connections to the IoP with higher probability than to other peers, thus resulting in significant performance improvements. IoP has been found to achieve a more efficient operation of the underlying network and thus a cost reduction for the ISPs together with performance improvements for the overlay users. Moreover, the combination of the IoP insertion together with locality awareness has been found to result in even further improvements of inter-domain traffic.

The purpose of the COMEBACK project (group internal research project) is to investigate the practical merit of theoretically established congestion control algorithms, designed for the transfer of time-elastic background data. According to their theoretical analysis and simulated experiments, these algorithms achieve to minimize the average delay of the interactive time-sensitive traffic traversing the same path, while providing a targeted throughput to the background transmissions that satisfies the relevant demand.

In the context of the project, a demo system was delivered including the development of the algorithms and the software for the quantification of their performance (C programming language, Ubuntu v16.04 Operating System). For the generation of the web (time-sensitive) IP traffic, the Apache JMeter application was utilized, targeting to closely capture the actual network conditions (e.g., the number of parallel TCP connections used for the transmission of the contents of a web page). The experimental performance evaluation of the algorithms was performed over a small-scale network, consisting by two bottleneck links. The findings indicate an alignment with the theoretical results, and particularly the algorithms achieve to decrease the interactive delay up to 35%, as the load of the interactive and background traffic tends to saturate the link’s capacity. Finally, the findings indicate that the algorithms clearly outperform the Less than Best Effort (e.g., TCP-LP) congestion control algorithms (when used for the background transfers), which are designed with the same objective, i.e., to prioritize the transmission of the delay-sensitive traffic.