Mobile Edge Computing in 5G Network

Modern cloud computing services are solely hosted by data centers and incapable of efficiently executing mobile applications due to the following reasons. First, several mobile applications require immediate response, and hence suffer from the excessive network latency accessing the remote data centers. Second, data centers provide virtualized cloud resources as Virtual Machines (VMs), each of which serves an enterprise user with high volumes of workloads or responds to a type of web requests. Should the approach be adopted to handle each mobile application, data centers will incur significant overhead for global VM provisioning and management. Edge computing is becoming the viable solution to reduce latency and overheads by placing VMs close to the mobile users, at the edge of the access network.

Recent researches in this area are the following.

Multi-slices MEC resources orchestration

The 5G mobile network will rely on network slicing to deal with a wide variety of services with different Quality of Service (QoS) requirements. Network slicing is promoted by 3GPP and provides a logical vertical partition of the network based on network virtualization technologies, i.e. Network Function Virtualization (NFV) and Software Defined Networking (SDN). Despite the undisputed benefits in terms of flexibility and scalability pledged by the paradigm, network slicing comes together with a novel and growing complexity mainly related to the planning and optimization of network resources. The spatio-temporal dynamics of the traffic demand relying on slices causes continuous expansion/shrinking of the resources needed to meet the requirements, thus making their orchestration more complicated. The research deals with the problem of providing an optimized plan for resources allocation at the edge of a sliced network. We model a scenario of two network slices which have different delay requirements and we model their traffic demand by exploiting an anonymized CDR dataset.
We provide an extensive analysis of the QoS and the user’s Quality of Experience (QoE), we show the beneficial impact the approach has on both mobile operators and their users, and we highlight the performance growth achieved versus a single slice approach of undifferentiated traffic.

Platoon autonomous driving through 5G network

Vehicles on the road with some common interests can cooperatively form a platoon-based driving pattern, in which a vehicle follows another vehicle and maintains a small and nearly constant distance to the preceding vehicle. It has been proved that, compared with driving individually, such a platoon-based driving pattern can significantly improve road capacity and energy efficiency. Moreover, with the emerging vehicular ad hoc network (VANET), the performance of a platoon in terms of road capacity, safety, energy efficiency, etc., can be further improved. On the other hand, the physical dynamics of vehicles inside the platoon can also affect the performance of the underlying network. Such a complex system can be considered a platoon-based vehicular cyber-physical system (VCPS), and has attracted significant attention recently. Our research aims to control the autonomous driving of platoons through the 5G network infrastructure.

Latest Selected Publications

Christian Quadri; Sabrina Gaito; Gian Paolo Rossi, Big data inspired, proximity-aware 4G/5G service supporting urban social interactions, in IEEE International Conference on Smart Computing (SMARTCOMP 2016), IEEE 2016.

Quadri, Christian; Gaito, Sabrina; Rossi, Gian Paolo, Proximity-aware offloading of person-to-person communications in LTE networks, in 13th IEEE Annual Consumer Communications & Networking Conference (CCNC), pp. 608–613, IEEE 2016.

Nika, Ana; Ismail, Asad; Zhao, Ben; Gaito, Sabrina; Rossi, Gian Paolo; Zheng, Haitao, Understanding and Predicting Data Hotspots in Cellular NetworksMobile Networks and Applications, pp. 1–12, 2015.