Bhattacharya, Ayushman, Chakraborty, Anwesha · Nov 1, 2026
Bhattacharya and Chakraborty’s arXiv paper argues guarded recovery can prevent harmful task migrations
Markdown · JSON · Original source
Ayushman Bhattacharya and Anwesha Chakraborty present QFabric, a prototype for recovering a task when it violates an empirical soft real- contract on a heterogeneous embedded board. The system compiles a bounded task for both the Linux application processor and the real-time microcontroller (MCU), then permits a move only after semantic and capacity admission at a quiescent epoch boundary. It monitors performance with hysteresis and can commit or roll back a move after probation. On an Arduino UNO Q running Linux and Zephyr, the authors measured five workloads under different contention conditions. They report that each MCU invocation adds about 19 ms of dispatch, and Linux contention can inflate the MCU path substantially because every call traverses Linux userspace. In their four-node-chain tests, Linux met every contract, so QFabric did not migrate; a controller without hysteresis migrated in seven cells and raised p95 by up to 177x, while probation rolled back five moves. The authors argue that on Linux-dispatched heterogeneous boards, migration cannot isolate a task from Linux interference, making recovery guards useful chiefly because they prevent harmful moves. They also say the paper supersedes an earlier artifact release whose reported improvement resulted from measurement defects. The source, raw traces, and generators are public.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
eess.SY · Article
Source details
Original source date: Nov 1, 2026 (2026-11-01).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-10T08:47:31.061Z. First observed: 2026-11-03T08:58:42.848Z. Retrieved: 2026-11-03T22:16:10.024Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
The reported measurements use an Arduino UNO Q with a Cortex-A53 Linux processor and a Cortex-M33 running Zephyr; the abstract does not establish how results transfer to other boards or workloads. In the reported four-node-chain tests, Linux met every contract, so the abstract does not report QFabric recovering from an actual contract violation. The abstract identifies measurement defects in an earlier artifact release but does not describe those defects in the supplied text.
Open question
What results would QFabric produce when Linux fails to meet a task's contract, given that the reported four-node-chain tests found Linux met every contract?
Source-stated
The authors present QFabric as guarded recovery for tasks that violate empirical soft real-time contracts, rather than as automatic cross-domain placement.
a prototype that treats cross-domain placement as guarded recovery from a violated empirical soft real-time contract
Source-stated
The prototype permits a switch only after semantic and capacity admission at a quiescent epoch boundary, and can commit or roll back after probation.
switched only after semantic and capacity admission at a quiescent epoch boundary, and committed or rolled back after probation
Source-stated
In the tested setup, every MCU invocation added about 19 ms of dispatch, and Linux contention inflated the MCU path more than the Linux path.
Each MCU invocation costs about 19 ms of dispatch regardless of compute
Source-stated
Across 30 cells of a four-node chain, Linux met every contract within 0.8% misses, so QFabric never migrated; a controller without hysteresis migrated in seven cells and raised p95 by up to 177x.
Across 30 cells of a four-node chain, Linux meets every contract within 0.8% misses and QFabric correctly never migrates; a controller without hysteresis migrates in 7 cells, raising p95 by up to 177x
Source-stated
The authors conclude that migration on Linux-dispatched heterogeneous boards cannot isolate a task from Linux interference, and that recovery guards mainly prevent harmful moves.
On Linux-dispatched heterogeneous boards, migration therefore cannot isolate a task from Linux interference, and recovery guards matter chiefly by preventing harmful moves.
Source-stated
The authors say this paper supersedes an earlier artifact release because its reported improvement arose from measurement defects.
This paper supersedes an earlier artifact release whose reported improvement arose from measurement defects that we identify.
Source and contributor context
Bhattacharya, Ayushman, Chakraborty, Anwesha · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: 6c072682-25cc-4f23-803c-7164335fbbe8.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-10T08:47:31.061Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: dfe1d7116e40baa3b1630cea341fc3183c378b98120d8626ddec3d9a5c389008. Source text hash: 282be8ab1b990d3e8ee372700609cd9de48cb7643f3a03fe7177b439a0ff6a2c.
Revision and source-check history
Version 1 · 2026-11-10T08:47:31.061Z
Bhattacharya and Chakraborty’s arXiv paper argues guarded recovery can prevent harmful task migrations
Ayushman Bhattacharya and Anwesha Chakraborty present QFabric, a prototype for recovering a task when it violates an empirical soft real- contract on a heterogeneous embedded board. The system compiles a bounded task for both the Linux application processor and the real-time microcontroller (MCU), then permits a move only after semantic and capacity admission at a quiescent epoch boundary. It monitors performance with hysteresis and can commit or roll back a move after probation. On an Arduino UNO Q running Linux and Zephyr, the authors measured five workloads under different contention conditions. They report that each MCU invocation adds about 19 ms of dispatch, and Linux contention can inflate the MCU path substantially because every call traverses Linux userspace. In their four-node-chain tests, Linux met every contract, so QFabric did not migrate; a controller without hysteresis migrated in seven cells and raised p95 by up to 177x, while probation rolled back five moves. The authors argue that on Linux-dispatched heterogeneous boards, migration cannot isolate a task from Linux interference, making recovery guards useful chiefly because they prevent harmful moves. They also say the paper supersedes an earlier artifact release whose reported improvement resulted from measurement defects. The source, raw traces, and generators are public.
Original source
eess.SY · Article
Source details
Original source date: Nov 1, 2026 (2026-11-01).
Date basis: Date basis not provided.
Added to Escape Meridian: 2026-11-10T08:47:31.061Z. First observed: 2026-11-03T08:58:42.848Z. Retrieved: 2026-11-03T22:16:10.024Z.
Limits
The reported measurements use an Arduino UNO Q with a Cortex-A53 Linux processor and a Cortex-M33 running Zephyr; the abstract does not establish how results transfer to other boards or workloads. In the reported four-node-chain tests, Linux met every contract, so the abstract does not report QFabric recovering from an actual contract violation. The abstract identifies measurement defects in an earlier artifact release but does not describe those defects in the supplied text.
Open question
What results would QFabric produce when Linux fails to meet a task's contract, given that the reported four-node-chain tests found Linux met every contract?
Source-stated
The authors present QFabric as guarded recovery for tasks that violate empirical soft real-time contracts, rather than as automatic cross-domain placement.
a prototype that treats cross-domain placement as guarded recovery from a violated empirical soft real-time contract
Source-stated
The prototype permits a switch only after semantic and capacity admission at a quiescent epoch boundary, and can commit or roll back after probation.
switched only after semantic and capacity admission at a quiescent epoch boundary, and committed or rolled back after probation
Source-stated
In the tested setup, every MCU invocation added about 19 ms of dispatch, and Linux contention inflated the MCU path more than the Linux path.
Each MCU invocation costs about 19 ms of dispatch regardless of compute
Source-stated
Across 30 cells of a four-node chain, Linux met every contract within 0.8% misses, so QFabric never migrated; a controller without hysteresis migrated in seven cells and raised p95 by up to 177x.
Across 30 cells of a four-node chain, Linux meets every contract within 0.8% misses and QFabric correctly never migrates; a controller without hysteresis migrates in 7 cells, raising p95 by up to 177x
Source-stated
The authors conclude that migration on Linux-dispatched heterogeneous boards cannot isolate a task from Linux interference, and that recovery guards mainly prevent harmful moves.
On Linux-dispatched heterogeneous boards, migration therefore cannot isolate a task from Linux interference, and recovery guards matter chiefly by preventing harmful moves.
Source-stated
The authors say this paper supersedes an earlier artifact release because its reported improvement arose from measurement defects.
This paper supersedes an earlier artifact release whose reported improvement arose from measurement defects that we identify.
Source and contributor context
Bhattacharya, Ayushman, Chakraborty, Anwesha · Article author
These are credits stated by the source; they do not establish a profile match or authorship of every passage.
Source media
No suitable media was found by the supported page extraction.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
arXiv
arXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: 6c072682-25cc-4f23-803c-7164335fbbe8.
Prepared by gpt-6-luna; grounding review by gpt-6-luna.
Source check: uncertain by anthropic/claude-opus-5.5 on 2026-11-10T08:47:31.061Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: material_development.
Observation: dfe1d7116e40baa3b1630cea341fc3183c378b98120d8626ddec3d9a5c389008. Source text hash: 282be8ab1b990d3e8ee372700609cd9de48cb7643f3a03fe7177b439a0ff6a2c.
uncertain · 2026-11-10T08:47:31.061Z · anthropic/claude-opus-5.5
Publication: 6c072682-25cc-4f23-803c-7164335fbbe8
attribution — pass: The paper, its claims and conclusions are attributed to authors Ayushman Bhattacharya and Anwesha Chakraborty, matching the arXiv page. Results are framed as the authors' reports.
Authors:Ayushman Bhattacharya, Anwesha Chakraborty
summary — uncertain: The reading tracks the abstract faithfully, including the supersession note. But central terms are left unexplained for a general reader: empirical soft real-time contract, hysteresis, quiescent epoch boundary, semantic and capacity admission, probation, p95, cells and four-node chain. Without these, the 177x result and why guards help are hard to follow. Smallest repair: add brief plain-language glosses or tooltips for hysteresis and p95. Optionally restore the 34x MCU slowdown and the p99 400 ms versus 17.0 ms comparison, which give the interference argument its force.
monitored in windows with hysteresis, switched only after semantic and capacity admission at a quiescent epoch boundary
date — pass: originalPublished 2026-11-01 matches the v1 submission date on arXiv. The feed date of Nov 3 is later and correctly not used.
[v1] Sun, 1 Nov 2026 05:28:32 UTC (28 KB)
support — pass: All six claims are supported by the abstract. The phrase in claim 1, rather than as automatic cross-domain placement, is a mild framing. It is consistent with the source treating placement as guarded recovery. The question and limitations follow from the reported results and do not add unsupported benefits or costs.
Linux contention inflates the MCU path far more than the Linux path (p99 up to 400 ms versus 17.0 ms)
media — pass: The selected media is the verified direct arXiv abstract page for this paper, which is the most direct access to the source.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMEarXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXivTIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMEarXiv is a research manuscript repository where authors share papers that may not yet have been peer reviewed.
Explore arXiv