Page 70 - ITU Journal Future and evolving technologies Volume 2 (2021), Issue 6 – Wireless communication systems in beyond 5G era
P. 70
ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 6
InP InP 1 InP 2 ... InP i
InP ... ...
MVNO 1 MVNO h MVNO 1 MVNO 2 MVNO h
... ...
MVNO 1 MVNO h SP 1 SP 2 SP k
SP 1 ... SP k
URLLC USERS URLLC USERS URLLC USERS URLLC USERS URLLC USERS URLLC USERS URLLC USERS
eMBB USERS eMBB USERS eMBB USERS eMBB USERS eMBB USERS eMBB USERS eMBB USERS
mMTC USERS mMTC USERS mMTC USERS mMTC USERS mMTC USERS mMTC USERS mMTC USERS
(a) Traditional Multi‑Tenancy (b) Single‑Domain Multi‑Tenant NS (c) Multi‑Domain Multi‑Tenant NS
Fig. 1 – Illustration of the domain type of network slicing.
lexibility of the network architecture. In an M‑TTSD net‑ packet loss probability, delay bound, slice users
work, an SP satis ies the demands of slice users by bidding distribution, cell load, tier load, and bidding budgets
for virtual resources from multiple MVNOs. Similarly, an of network entities such as MVNO and SP) and slice
MVNO bids for mobile network resources from multiple users’ parameters (such as slice users’ location, slice
InPs to meet the demands of SPs associated with it. The use case QoS requirement, associated interference).
M‑TTSD network explores the bene its and possibilities of
tenants connecting to several network domains for cover‑ 2. Other than the traditional two‑player network ap‑
age extension, scalability, and network resource optimi‑ proach, we consider a multi‑tenant multi‑domain
sation. network with entities comprising InPs, MVNOs, and
Ef icient resource management is pivotal to the opti‑ SPs, respectively. To this end, a three‑stage multi‑
mal operation of the M‑TTSD network [5, 6]. More‑ domain auction game based on the Fisher Market
over, static slicing and centralised resource management (FM) principle and shared‑constrained proportion‑
frameworks would not be practicable in M‑TTSD net‑ ality is exploited to facilitate an agile and dynamic
works. To fully exploit the lexibility and dynamic char‑ business model for 5G NS and beyond networks; and
acterisation of the slice traf ic, it is important to note also to maximise the utility of the respective players.
that not adequately addressing the resource allocation The InPs, MVNOs, and SPs trade network resources
challenge in an M‑TTSD network will adversely affect the to meet the demands of slice users in a manner that
Quality‑of‑Service (QoS) of slice users and the IC of net‑ ICs and Individual Rationality (IR) are not compro‑
mised.
work players, thereby jeopardising the sustainability of
5G slice networks and future networks. 3. We formulate the slice users’ service selection prob‑
The main contributions of this work can be summarised lem in an M‑TTSD network as a maximisation prob‑
as follows. lem. To reduce the complexity involved in solving
the formulated problem, a hierarchical decomposi‑
1.1 Contributions tion technique is employed. We develop a multistage
matching‑theory inspired scheme to optimally asso‑
The main contributions of this work can be summarised ciate slice users to SPs, MVNOs, and InPs, respec‑
as follows. tively, in an M‑TTSD network. The multistage match‑
ing algorithm considers the Signal‑to‑Interference‑
1. We consider a latency‑aware dynamic resource plus‑Noise Ratio (SINR) of the slice users, slice QoS
allocation framework for an M‑TTSD network with requirements in the course of matching slice users
enhanced Mobile Broadband (eMBB), massive to the respective network tiers and InP.
Machine‑Type Communications (mMTC), and Ultra‑
Reliable Low‑Latency Communications (URLLC) 4. We develop a distributed backtracking algorithm
slice users, respectively. The framework allocates that aids buyers and sellers in the respective stages
resources to slice users by engaging network pa‑ to trade network resources in an incomplete infor‑
rameters (such as packet size, packet arrival rate, mation scenario. The backtracking algorithm takes
58 © International Telecommunication Union, 2021