Dynamic Preamble Subset Allocation for RAN Slicing in 5G Networks

Abstract

The random access (RA) mechanism of Long Term Evolution (LTE) networks is prone to congestion when a large number of devices attempt RA simultaneously, due to the limited set of preambles. If each RA attempt is made by means of transmission of multiple consecutive preambles (codewords) picked from a subset of preambles, as proposed, collision probability can be significantly reduced. Selection of an optimal preamble set size can maximize RA success probability in the presence of a trade-off between codeword ambiguity and code collision probability, depending on load conditions. In light of this finding, this paper provides an adaptive algorithm, called Multi-preamble RA , to dynamically determine the preamble set size in different load conditions, using only the minimum necessary uplink resources. This provides high RA success probability, and makes it possible to isolate different network service classes by separating the whole preamble set into subsets each associated to a different service class; a technique that cannot be applied effectively in LTE due to increased collision probability. This motivates the idea that preamble allocation could be implemented as a virtual network function, called vPreamble, as part of a random access network slice. The parameters of a vPreamble instance can be configured and modified according to the load conditions of the service class it is associated to.