Paper Info
Title
End-to-End Millimeter-Wave Network Performance and Mobility Management Overhead in Urban Cellular Deployments with Realistic Pedestrian Traffic and Blockages
Abstract
Millimeter-wave (mm-wave) frequencies are seen as the new frontier in enabling high capacity for 5G-and-beyond cellular networks. However, spatio-temporal variations of the mm-wave channel impose significant challenges for high rate error-free communication, due to increased propagation loss and blockage susceptibility. In this paper, we analyze the end-to-end mobile user performance and the extent of mobility management overhead in the presence of dynamic pedestrian blockages. We use site-specific propagation data obtained via high resolution ray-tracing and realistic pedestrian traffic to implement site-specific user mobility and pedestrian blockage models, and integrate these in an extensive ns-3 framework to study the end-to-end mobile user performance. Our results show that the number of mm-wave mobility management events -- beam tracking, beam steering and handover -- strongly depends on the pedestrian walk nature and number of blockage events, i.e. the primary mm-wave serving link can be blocked for up to 36% of a walk duration. These frequent short-term pedestrian blockages can lead to TCP throughput degradation of up to 35% and TCP round-trip-time spikes of over 1 s, compared to the achievable network performance when pedestrian blockages are not considered, assuming mobility management with up-to-date CSI. We also show that stale CSI -- arising from delay in periodic beam training procedures -- results in more frequent and sub-optimal mobility management decisions and an additional TCP throughput degradation of up to 30%, compared to the achievable TCP throughput assuming mobility management with up-to-date CSI.
Year
DOI
Venue
2020
10.1145/3416012.3424621
MSWiM '20: 23rd International ACM Conference on Modeling, Analysis and Simulation of Wireless and Mobile Systems Alicante Spain November, 2020
DocType
ISBN
Citations 
Conference
978-1-4503-8119-2
1
PageRank 
References 
Authors
0.37
0
3
Name
Order
Citations
PageRank
Aleksandar Ichkov175.55
Petri Mahonen233035.46
Ljiljana Simic310917.17