|AKGÜL ÖZGÜR UMUT||Cycle: XXXI |
Tutor: CESANA MATTEO Abstract:
Performance and enablers of infrastructure and resource sharing
As a reaction to the constraints and inabilities in 4G, improved spectral efficiency, high coverage area, reduced latency (< 1ms latency), ultra-high data rates and capacity and large energy savings as well as cost reduction are some of the promises of 5G technology. However, these promises of 5G technology also reveal bold challenges. First of all, the trade-o? between cost e?ciency and service utilization is a crucial problem in 5G networks. Since the wireless traffic is saturating quickly, the network operators have to provide economical services to maintain both stability of their business framework and their customer profile. With the introduction of Internet of Things (IoT) on the top of huge user and sensor network, the operators obliged to buy huge amounts of infrastructure resources. However, during the infrastructure purchase phase, the network operators are incapable of determining the future demands of the users. Being blind to determine the needs of the network, the network operators purchase capacity that will possibly mismatched to their needs. If the purchased capacity is lower than the actual demand then the network operator will be incapable of providing service to some of his customers. If the capacity is higher than the actual need then the network operator will be paying for the infrastructure that is not used.
From a network operator¿s perspective, sharing degrades the critically of the capacity purchasing process and lets the network operators to decrease their capital costs. Moreover, since the network operators can share their unused capacities, the operational costs will be shared between multiple operators. Besides, the network operators would be able to use the excess capacity of the other network operators in times of bursty tra?c. Another major proposal of sharing virtualized infrastructure is evolvability. Virtualized systems divide the physical infrastructure resources into virtual resource blocks and give a mapping between these resource blocks and the physical resources. While the network controller makes changes in one of the virtualized resource blocks, the network operator would be capable of continuing his service using other virtual blocks. Finally, infrastructure sharing decreases the risk of new market entry as it allows the new entrants to hire some capacity for certain amount of time. Through this time, the new operator would be able to have an exact insight of the customer responses. Moreover, by sharing the capital costs of rural coverage, the operators can reduce their risk of covering a low business potential area. In spite of these advantages, the number of existing sharing agreements are quite low.
The research on the feasibility of network sharing is performed through the PhD research activities. In the ?rst part, the economic and secondary motivations of di?erent sharing options will be investigated. Through the researches so far, it is observed that for short term sharing agreements, the economic gains of the network operators govern the decision of infrastructure sharing. On the other hand, for long term agreements, secondary metrics become dominant and shape the sharing decision. In the literature, to the best of our knowledge, the radio access network (RAN) sharing is the dominant sharing option for long term sharing agreements. This is mostly because the other type of RAN sharing approach, e.g. capacity sharing, is a type of national roaming and this action is usually de?ned by service level agreements (SLAs) rather than technical analysis. Following the analysis on the network operator¿s intent on sharing, enquiries will be focused on the applicability of the sharing in terms of technical necessities. More specifically, assuming that the sharing is economically preferable and the network operators have an intent to share their resources, the technical models as well as the possible resources to share will be investigated. As a generic approach, the network operators can share their spectrum or their capacity. The algorithmic viability and fairness are the parts to be investigated in this phase. After the analysis on both the economical and technical applicability of the network sharing, the resources will be continued on the necessary virtualization technology to perform the approaches. As a starting point, the software defined networks and the network function virtualization technologies will be investigated. Their strengths and weaknesses will be observed for the proposed algorithms. In case of necessity, alternative technologies and their applications will be researched.
Advisor: Prof. Antonio Capone