• Melisa Lopez Lechuga
4. term, Wireless Communication Systems, Master (Master Programme)
The 5th Generation (5G) New Radio (NR) air interface is expected to be the foundation of very heterogeneous networks serving a wide range of use cases, including Ultra-Reliable Low Latency Communications (URLLC) services. In URLLC, small data packets must be correctly transmitted and received in a short time with high reliability (up to 1 ms latency with a success probability of 99.999%). Different options are being considered to meet this challenging design target. One such considered solution is data duplication through dual connectivity, where the same packet is independently transmitted through two different nodes.
This project studies the functionality of data duplication at PDCP level for dual connectivity through system-level simulations, where each packet copy is sent through the two links to which a certain UE is connected. The studied scenario is a heterogeneous network of 21 macro cells with a cluster of 4 pico cells per macro cell area. The scenario is first optimized for the single connectivity case, which supports up to 8Mbps URLLC load while meeting the URLLC requirements. When dual connectivity is enabled, in a controlled manner, in a URLLC traffic only scenario, it is shown that dual connectivity does not provide any gain due to the low interference conditions. As second step, the benefit of DC is studied when the URLLC traffic coexist with full buffer background eMBB traffic. Results show that latency gain can be obtained by dual connectivity, however the sensitivity of this gain on the scenario conditions is quite high. Finally, an optimization is added, in which if a packet sent through one of the links is successfully received at the UE, the transmission of its copy on the other link is cancelled (i.e. the packet is discarded at the network side). This optimization results in a performance improvement in terms of the latency especially at high load because it avoids buffering delay.
Publication date3 May 2018
Number of pages63
ID: 274597457