Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Analysis of Age of Information threshold violations

Franco, Antonio LU ; Landfeldt, Björn LU and Körner, Ulf LU (2019) 22nd ACM International Conference on Modeling, Analysis
and Simulation of Wireless and Mobile Systems
Abstract
We study a scenario where a monitor is interested in the freshest possible update from a remote sensor. The monitor also seeks to minimize the number of updates that exceed a certain freshness threshold, beyond which, the information is deemed to be too old. Previous work has presented results for First Come First Served (FCFS) systems. However, it has been shown that Last Come First Served (LCFS) with preemption is more effective in terms of average Age of Information (AoI); we therefore study an M/G/1 LCFS system with preemption. The generality of the busy time distribution gives the advantage of applicability on any distribution inside the model. For example, one can use a deterministic distribution to study a TDMA system, a gamma... (More)
We study a scenario where a monitor is interested in the freshest possible update from a remote sensor. The monitor also seeks to minimize the number of updates that exceed a certain freshness threshold, beyond which, the information is deemed to be too old. Previous work has presented results for First Come First Served (FCFS) systems. However, it has been shown that Last Come First Served (LCFS) with preemption is more effective in terms of average Age of Information (AoI); we therefore study an M/G/1 LCFS system with preemption. The generality of the busy time distribution gives the advantage of applicability on any distribution inside the model. For example, one can use a deterministic distribution to study a TDMA system, a gamma distribution to model a routing network, or a more complicated distribution to study a CSMA access scheme. We find a general procedure to derive the exact expression of the outage update probability -- i.e. the portion of time updates have information older than a certain threshold. We compare different busy time distributions to the ones already present in literature for equivalent FCFS systems, showing the benefit of using the former discipline. We further study how the variance of the busy time distribution affects the update outage probability. We compare the M/D/1 LCFS with preemption against the M/$\Gamma$/1 LCFS with preemption and let the variance of the busy time of the latter vary, while maintaining the same average busy time for both systems. We find that at low thresholds and low loads, higher variance gives an advantage in terms of update outage probability. (Less)
Abstract (Swedish)
We study a scenario where a monitor is interested in the freshest possible update from a remote sensor. The monitor also seeks to minimize the number of updates that exceed a certain freshness threshold, beyond which, the information is deemed to be too old. Previous work has presented results for First Come First Served (FCFS) systems. However, it has been shown that Last Come First Served (LCFS) with preemption is more effective in terms of average Age of Information (AoI); we therefore study an M/G/1 LCFS system with preemption. The generality of the busy time distribution gives the advantage of applicability on any distribution inside the model. For example, one can use a deterministic distribution to study a TDMA system, a gamma... (More)
We study a scenario where a monitor is interested in the freshest possible update from a remote sensor. The monitor also seeks to minimize the number of updates that exceed a certain freshness threshold, beyond which, the information is deemed to be too old. Previous work has presented results for First Come First Served (FCFS) systems. However, it has been shown that Last Come First Served (LCFS) with preemption is more effective in terms of average Age of Information (AoI); we therefore study an M/G/1 LCFS system with preemption. The generality of the busy time distribution gives the advantage of applicability on any distribution inside the model. For example, one can use a deterministic distribution to study a TDMA system, a gamma distribution to model a routing network, or a more complicated distribution to study a CSMA access scheme. We find a general procedure to derive the exact expression of the outage update probability -- i.e. the portion of time updates have information older than a certain threshold. We compare different busy time distributions to the ones already present in literature for equivalent FCFS systems, showing the benefit of using the former discipline. We further study how the variance of the busy time distribution affects the update outage probability. We compare the M/D/1 LCFS with preemption against the M/Γ/1 LCFS with preemption and let the variance of the busy time of the latter vary, while maintaining the same average busy time for both systems. We find that at low thresholds and low loads, higher variance gives an advantage in terms of update outage probability. (Less)
Please use this url to cite or link to this publication:
author
; and
organization
alternative title
Analys av överskridna tröskelvärden av informationsålder
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
host publication
ACM International Conference on Modeling, Analysis and Simulation of Wireless and Mobile Systems
pages
10 pages
publisher
Association for Computing Machinery (ACM)
conference name
22nd ACM International Conference on Modeling, Analysis <br/>and Simulation of Wireless and Mobile Systems
conference location
Miami Beach, United States
conference dates
2019-11-25 - 2019-11-29
external identifiers
  • scopus:85077320623
ISBN
978-1-4503-6904-6
DOI
10.1145/3345768.3355909
language
English
LU publication?
yes
id
6c572498-ee72-4a4d-bc8d-7db07b8dc240
date added to LUP
2019-09-02 09:53:10
date last changed
2022-05-04 00:45:53
@inproceedings{6c572498-ee72-4a4d-bc8d-7db07b8dc240,
  abstract     = {{We study a scenario where a monitor is interested in the freshest possible update from a remote sensor. The monitor also seeks to minimize the number of updates that exceed a certain freshness threshold, beyond which, the information is deemed to be too old. Previous work has presented results for First Come First Served (FCFS) systems. However, it has been shown that Last Come First Served (LCFS) with preemption is more effective in terms of average Age of Information (AoI); we therefore study an M/G/1 LCFS system with preemption. The generality of the busy time distribution gives the advantage of applicability on any distribution inside the model. For example, one can use a deterministic distribution to study a TDMA system, a gamma distribution to model a routing network, or a more complicated distribution to study a CSMA access scheme. We find a general procedure to derive the exact expression of the outage update probability -- i.e. the portion of time updates have information older than a certain threshold. We compare different busy time distributions to the ones already present in literature for equivalent FCFS systems, showing the benefit of using the former discipline. We further study how the variance of the busy time distribution affects the update outage probability. We compare the M/D/1 LCFS with preemption against the M/$\Gamma$/1 LCFS with preemption and let the variance of the busy time of the latter vary, while maintaining the same average busy time for both systems. We find that at low thresholds and low loads, higher variance gives an advantage in terms of update outage probability.}},
  author       = {{Franco, Antonio and Landfeldt, Björn and Körner, Ulf}},
  booktitle    = {{ACM International Conference on Modeling, Analysis  and Simulation of Wireless and Mobile Systems}},
  isbn         = {{978-1-4503-6904-6}},
  language     = {{eng}},
  month        = {{11}},
  publisher    = {{Association for Computing Machinery (ACM)}},
  title        = {{Analysis of Age of Information threshold violations}},
  url          = {{https://lup.lub.lu.se/search/files/69041667/mswim06.pdf}},
  doi          = {{10.1145/3345768.3355909}},
  year         = {{2019}},
}