QoS-Aware Radio Access Technology (RAT) Selection in Hybrid Vehicular Networks

Enrique Alba; Fethi Filali; Jamal Toutouh; Zeeshan Hameed Mir

REVIEW 4 major objections 5 minor 15 references

QoS-Aware Radio Access Technology (RAT) Selection in Hybrid Vehicular Networks

T0 review · 4 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read When load rises, the proposed algorithm slows beaconing before switching to LTE, cutting vertical handovers and improving PDR, latency, and goodput.

desk verdict A sensible but overclaimed workshop paper: the BFA-then-VHO mechanism is clear and directionally supported, yet the abstract promises significant PDR/goodput gains that the text itself says are comparable. read the letter →

arxiv 2501.12304 v1 pith:4HRS2PZZ submitted 2025-01-21 cs.NI

Zeeshan Hameed Mir , Jamal Toutouh , Fethi Filali , Enrique Alba This is my paper · ORCID

classification cs.NI

keywords RATselectionHybridvehicularnetworksIEEE802.11pLTEVerticalhandoverBeaconingfrequencyadaptationQoSNetworkload

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that in a hybrid vehicular network combining IEEE 802.11p ad hoc links with LTE cellular access, the right response to congestion is not to switch radio technology immediately but to first slow down the rate at which vehicles broadcast beacons. The proposed QoS-aware RAT selection algorithm monitors the 802.11p network load; when load crosses a threshold, it reduces the beaconing frequency locally, then asks neighboring vehicles to do the same, and only performs a vertical handover to LTE if load remains high. The authors show by simulation that this staged mechanism produces fewer vertical handovers than periodic RAT selection and than switching on load alone, while improving packet delivery ratio, latency, and application-level throughput. The contribution matters because vertical handovers are costly in signaling and delay, and hybrid vehicular networks need to keep safety and infotainment traffic flowing with minimal disruption.

What carries the argument

The load-bearing mechanism is Beaconing Frequency Adaptation (BFA), a control loop that reduces an application's beacon transmission rate in graded steps defined by a reduction factor (rFactor) up to a maximum tolerance (rTolerance), with two timers governing how long the reduced rate may hold (T(bFreqReduced)) and how long the initial rate must be restored afterward (T(bFreqInitial)). BFA is triggered by a Network Load Monitor (NLM) when the 802.11p queue length exceeds a threshold (NLM-Threshold, 80–90% of queue capacity). The central object is the staged decision rule: local BFA first, cooperative BFA among neighbors second, vertical handover to LTE only as a last resort. This ordering is what converts a load spike into a brief rate reduction instead of a costly handover, and it is the mechanism the paper credits for the reduced VHO count and improved performance metrics.

What would settle it

Re-run the same 1 km, three-lane, 150-vehicle highway scenario in a standards-faithful simulation that models 802.11p CSMA/CA contention and LTE handover signaling, and compare the proposed scheme against the same baselines; if the VHO reduction and PDR/latency/goodput improvements do not reproduce, the central claim fails.

Watch

Extended reading notes

Core claim

The paper's central claim is that a distributed radio resource manager can satisfy application QoS requirements in a hybrid IEEE 802.11p/LTE vehicular network by treating beaconing-frequency adaptation as a first-class tool before resorting to vertical handover. The algorithm defines an NLM-Threshold on the 802.11p queue lengths; when load exceeds it, the DRRM first applies BFA locally, then propagates the reduction request to neighboring vehicles, and only if that fails to bring load below threshold does it hand over traffic to LTE. Simulation results compare this mechanism against periodic RAT selection, a no-BFA load-triggered handover, and an 802.11p-only network, reporting fewer vertical handovers and higher packet delivery ratio, lower latency, and higher goodput for the proposed scheme.

Load-bearing premise

All performance conclusions come from a self-developed simulator whose 802.11p and LTE models are simplified—a single eNodeB, one 10 MHz cell, no interference or contention modeling—so the reported gains may not survive realistic radio conditions.

Editorial extensions

If this is right

If the staged BFA-then-VHO rule works, network operators can reduce load on the cellular interface, lowering the cost of vertical handovers and the data throughput lost during switching.
Applications with higher rTolerance (more tolerance for reduced beaconing) and longer T(bFreqReduced) will trigger fewer vertical handovers, so QoS-aware applications can be tuned to minimize cellular usage.
In the proposed scheme, the 802.11p interface carries the majority of beacons, while LTE serves only as an overflow; this suggests the algorithm can help keep cellular resources available for other traffic.
The comparison against periodic RAT selection implies that event-driven, load-triggered decisions with a local congestion-relief step dominate time-based proactive handovers in both handover count and QoS metrics.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The staged design suggests a general principle for any hybrid access network: before switching to a more expensive or scarce radio resource, first exploit the application's tolerance for degraded parameters on the current interface—here, beacon rate.
Because the simulation omits channel contention and LTE handover signaling, the quantitative gains in a real deployment are likely to be smaller; this is an inference, not a paper claim.
An obvious extension is coupling BFA with the 802.11p congestion-control mechanisms the paper cites, to test whether the two approaches complement each other or interfere.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

First thing you should know: this is a 2015 workshop paper that reads like one. The proposal is sensible — use beacon frequency adaptation to relieve local congestion before falling back to a load-triggered vertical handover to LTE — and the writing is clear enough that you can follow the algorithm and the simulation design. The comparative study against periodic RAT selection, No-BFA, and No-LTE is a reasonable baseline set. Where it falls down is the gap between the abstract and the results. The abstract promises “significant performance improvements” in PDR, latency, and throughput, but Section 4.2 says the PDR of No BFA and the proposed mechanism “are quite comparable” and that dual-interface schemes “attain comparable goodput.” Only latency is consistently better, plus the minimum (not median or maximum) VHO count. So the headline claim is internally contradicted and needs to be narrowed.

The evidence base is also thin. The simulator is self-developed, no code is released, no confidence intervals or statistical tests are reported, and the LTE model is one eNodeB with a single cell and no interference. The BFA equation in Section 3.2 is mangled — it reads “bFreqReduced = [bFreqReduced − (rFactor % × bFreqInitial)|” — and the worked example (8, 6, 4 Hz from a 10 Hz start with 25% steps) only works if you assume rounding to the nearest integer. That is the kind of detail a referee would want fixed. The parametric results on rTolerance and timers are mostly consequences of the parameter definitions, so they are sensitivity checks, not independent findings.

None of this kills the idea. The staged approach — first reduce beacons, then ask neighbors, then hand over — is a plausible and clearly articulated design, and the simulation results are directionally consistent with it. The paper deserves to be seen by a referee if it comes back for review, but the referee should require that the claims match the measured results, the simulator be validated against an established tool, and the statistics be reported properly. As it stands, treat it as a preliminary idea, not as evidence for significant throughput or reliability gains.

Who this is for: anyone working on RAT selection or congestion control in vehicular networks who wants a quick look at one way to combine BFA with VHO. It is not a landmark, but it is an honest attempt.

Referee Report

4 major / 5 minor

Summary. The paper proposes a QoS-aware Radio Access Technology (RAT) selection algorithm for hybrid vehicular networks that combine IEEE 802.11p ad hoc communication with LTE. A Distributed Radio Resource Management (DRRM) entity monitors IEEE 802.11p queue load; when congestion exceeds an NLM-Threshold, it first applies Beaconing Frequency Adaptation (BFA) and only performs a vertical handover (VHO) to LTE if local and neighbor-initiated frequency reductions cannot relieve the load. The paper reports simulations showing that the proposed mechanism reduces the number of VHOs and claims significant improvements in packet delivery ratio, latency, and application-level throughput compared with periodic RAT selection, a No-BFA load-based scheme, and a No-LTE 802.11p-only scheme.

Significance. If the claims were fully supported, the work would offer a practical, low-complexity heuristic for reducing unnecessary VHOs in hybrid vehicular networks, with the BFA-before-handover idea being a sensible and potentially useful design that aligns with existing channel congestion control approaches. The paper also makes a useful distinction between local and cooperative frequency adaptation. However, the current evidence base is a single self-developed, unvalidated simulator with no released code, no confidence intervals, and no statistical tests. The headline claim of significant improvements on all three QoS metrics is directly contradicted by the paper's own qualitative descriptions of the results. The central contribution is therefore plausible but not yet convincingly demonstrated.

major comments (4)

[§3.2 (Eq. 1) and worked example] The BFA equation is malformed and internally inconsistent. The printed expression 'bFreqReduced = [bFreqReduced − (rFactor % × bFreqInitial)|' contains unbalanced brackets, and the worked example does not follow from it. With bFreqInitial = 10 Hz and rFactor = 25% of bFreqInitial, a 25% reduction would give bFreqReduced = 7.5 Hz, 5.0 Hz, 2.5 Hz over the three iterations, not the stated 8 Hz, 6 Hz, 4 Hz. This is a load-bearing error because the BFA mechanism cannot be implemented or reproduced as specified.
[Abstract, §4.2, and Conclusion] The abstract and conclusion claim 'significant performance improvements in terms of packet delivery ratio, latency and application-level throughput,' but the comparative study in Section 4.2 states that PDR for No BFA and the proposed mechanism are 'quite comparable' (Fig. 7) and that schemes with dual-interfaces 'attain comparable goodput' (Fig. 9). Only latency is consistently improved, and the VHO reduction is reported for the minimum (not median or maximum) of the proposed scheme. The headline claim is thus internally inconsistent with the presented results and must be narrowed to 'fewer VHOs and lower latency' or be backed by statistical evidence of PDR/goodput gains.
[§4.1 and §4.2] All quantitative conclusions rest on a self-developed simulation tool that is not validated, not released, and analyzed only with means over 10 instances. The LTE model is a single 10 MHz cell with one eNodeB and no interference or backhaul model, and the 802.11p model uses a simplified propagation model with no contention modeling described. Given that the central performance claims depend entirely on this simulator, the paper should either validate against a standard simulator, provide error bars and statistical significance tests, or clearly discuss how each simplification could affect the reported PDR, latency, and goodput comparisons.
[§4.2, Fig. 4(b)] The observed decrease in VHOs with increasing rTolerance is essentially a consequence of the parameter's definition: rTolerance is the maximum tolerable beaconing-frequency reduction, so a larger value allows the BFA to reduce the frequency further before a handover is required. This is a parameter sensitivity result, not an emergent property of the algorithm. The paper should present Fig. 4 as a sensitivity analysis and avoid framing it as evidence of the algorithm's intrinsic superiority.

minor comments (5)

[Throughout] There are several typographical and grammatical errors, e.g., 'preformed' instead of 'performed' in the Introduction, 'Mhz' instead of 'MHz' in Section 4.1, and inconsistent capitalization of 'rFactor' (sometimes 'rfactor').
[§3.2] The notation in the BFA equation is confusing because bFreqReduced is used both as the target variable and as the running value, and the percentage sign placement is unclear. Please rewrite the equation with explicit iteration indexing, e.g., bFreqReduced^{(k+1)} = bFreqReduced^{(k)} − (rFactor/100) · bFreqInitial.
[§4.1] The term 'Three-Log Distance propagation model' is likely intended to be 'Three-Log-Distance' or 'Log-Distance'; please clarify the exact propagation model and the parameters used (e.g., path-loss exponents, shadowing).
[§4.2, Figs. 6–9] The figures show only summary statistics (e.g., min/median/max or stacked bars) without variance information. Adding confidence intervals or box plots would make the comparisons more informative and would allow the reader to assess whether the reported differences are meaningful.
[References] References [10] and [13] cite the same author group and are given with inconsistent formatting ('Anna et. al.' vs. 'Vegni, A.'); also, reference [15] appears to be a standard CRRM paper and should be cited with complete data.

Circularity Check

1 steps flagged · score 2.0 of 10

Minor definitional content in the rTolerance/VHO trend, but no load-bearing self-citation or fitted-input prediction; the main algorithm comparisons are independent simulation results.

self definitional [Section 3.2 (BFA parameters) and Section 4.2, Fig. 4(b)]
"The higher rTolerance values let applications to reduce the beaconing frequency at maximum which leads to lower network load and therefore less number of VHOs."

rTolerance is defined in Section 3.2 as the maximum tolerable reduction in beaconing frequency, and Step 3 performs a VHO only when 'the beaconing frequency cannot be further reduced without significantly sacrificing the QoS requirements.' Therefore, fewer VHOs with higher rTolerance follows directly from the algorithm's own definitions: a larger rTolerance mechanically enlarges the range of BFA reductions before the VHO condition is reached. The Fig. 4(b) observation is thus a restatement of the parameter's semantics rather than an independent empirical result.

full rationale

The central contribution is an algorithm plus an ablation-style simulation study, not a derivation from first principles. No parameter is fitted to data and then renamed as a prediction; no uniqueness theorem is imported from the authors' prior work; and the self-citations (refs [2] and [3]) merely motivate hybrid vehicular networking and are not load-bearing for the proposed mechanism. The only definitional element is the rTolerance/VHO trend: because BFA halts at rTolerance and VHO occurs only when further reduction is impossible, higher rTolerance must produce fewer VHOs by construction. That parametric observation is therefore not independent evidence, but it is not the paper's central claim. The comparisons against periodic RAT selection, No BFA, and No LTE are legitimate simulation benchmarks, and the reported latency and goodput behavior is not forced by the algorithm definitions. Separately, the paper's own Section 4.2 text says PDR for No BFA and the proposed mechanism are 'quite comparable' and that schemes with dual interfaces attain 'comparable goodput,' which undercuts the abstract's claim of 'significant performance improvements' in PDR and goodput; this is a correctness and support problem, not circularity. Overall circularity is minor and confined to the parametric impact study.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim depends on several hand-set thresholds and timer parameters, plus domain assumptions about queue-based congestion detection, beacon-rate QoS representation, and the fidelity of a simplified single-cell LTE model. No fundamentally new physical or algorithmic entity is introduced; DRRM is cited from prior work and BFA is a mechanism rather than an entity.

free parameters (5)

NLM-Threshold = 80%-90% of queue capacity
Used to decide when to trigger BFA and VHO; chosen by hand rather than derived, and the paper states it 'often varies' in this range.
rFactor = 25% in the example, varied in simulations
Controls the per-iteration reduction in beaconing frequency; its value directly affects the number of VHO steps.
rTolerance = 50% in the example, varied in simulations
Maximum tolerable frequency reduction as a percentage of initial frequency; determines how far BFA can go before VHO.
T(bFreqReduced) = Varied in simulations
Duration an application may operate at reduced beaconing frequency; longer values reduce VHOs in the simulations.
T(bFreqInitial) = Varied in simulations
Minimum duration at the initial frequency after a reduced period; longer values increase network load and VHOs.

assumptions (4)

domain assumption Queue length at the IEEE 802.11p interface is a valid indicator of congestion.
The NLM mechanism in Section 3.1 uses queue length thresholds to infer load; the paper does not validate that queue length correlates with delay or loss in the simulated channel.
domain assumption Application QoS can be fully characterized by bFreqInitial, rFactor, and rTolerance.
Section 3.2 assumes these three parameters capture QoS, with no model for packet loss, deadline, or safety-criticality beyond beacon rate.
domain assumption A simplified single-cell LTE model is representative of real LTE behavior.
Section 4.1 models one eNodeB, one 10 MHz cell at 900 MHz, and no inter-cell interference or detailed handover signaling, yet LTE latency and throughput are central to the comparison.
domain assumption The self-developed simulator accurately reproduces IEEE 802.11p and LTE performance.
All results come from this tool, described only briefly in Section 4.1, with no validation against established simulators or measurements.

how reviews work

0 comments

Cite this review

Pith. "Pith review of QoS-Aware Radio Access Technology (RAT) Selection in Hybrid Vehicular Networks." pith.science (2026). https://pith.science/paper/4HRS2PZZ

@misc{pith2026250112304,
  author       = {Pith},
  title        = {Pith review of: QoS-Aware Radio Access Technology (RAT) Selection in Hybrid Vehicular Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4HRS2PZZ}},
  note         = {Machine review of arXiv:2501.12304}
}

read the original abstract

The increasing number of wireless communication technologies and standards bring immense opportunities and challenges to provide seamless connectivity in Hybrid Vehicular Networks (HVNs). HVNs could not only enhance existing applications but could also spur an array of new services. However, due to sheer number of use cases and applications with diverse and stringent QoS performance requirements it is very critical to efficiently decide on which radio access technology (RAT) to select. In this paper a QoS aware RAT selection algorithm is proposed for HVN. The proposed algorithm switches between IEEE 802.11p based ad hoc network and LTE cellular network by considering network load and application's QoS requirements. The simulation-based studies show that the proposed RAT selection mechanism results in lower number of Vertical Handovers (VHOs) and significant performance improvements in terms of packet delivery ratio, latency and application-level throughput.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

15 extracted references · 15 canonical work pages

[1]

IEEE Transactions on Mobile Computing (TMC) 99 (2014)

Park, Y., Kuk, S., Kim, H., Ha, J., Liang, C.J., JeongGil, K.: A feasibility study and development framework design for realizing smartphone -based vehicular networking systems. IEEE Transactions on Mobile Computing (TMC) 99 (2014)

work page 2014
[2]

EURASIP Journal on Wireless Communications and Networking 2014(1), 89 (2014)

Hameed Mir, Z., Filali, F.: Lte and ieee 802.11p for vehicular networking: A performance evaluation. EURASIP Journal on Wireless Communications and Networking 2014(1), 89 (2014)

work page 2014
[3]

In: IEEE 79th Vehicular Technology Conference (VTC2014-Spring) (May 2014)

Hameed Mir, Z., Filali, F.: On the Performance Comparison between IEEE 802.11 p and LTE-based Vehicular Networks. In: IEEE 79th Vehicular Technology Conference (VTC2014-Spring) (May 2014)

work page 2014
[4]

Vinel, A.: 3gpp lte versus ieee 802.11 p/wave: which technology is able to support cooperative vehicular safety applications? IEEE Wireless Communications Letters 1(2), 125–128 (2012)

work page 2012
[5]

ISO/DIS 21217 (2006)

ISO: Intelligent transport systems-communications access for land mobiles (calm)architecture. ISO/DIS 21217 (2006)

work page 2006
[6]

IEEE Draft Standard P802.21/D13 (2008)

IEEE: Local and metropolitan area networks: Media independent handover services. IEEE Draft Standard P802.21/D13 (2008)

work page 2008
[7]

IEEE Std 802.11p-2010 (2010)

IEEE: Ieee standard for information technology –local and metropolitan area networks–specific requirements –part 11: Wireless lan (mac) and (phy) specifications amendment 6:wave. IEEE Std 802.11p-2010 (2010)

work page 2010
[8]

LTE: Overview of 3gpp release 8, http://www.3gpp.org/Release -8 (2010)

work page 2010

Show all 15 references

[9]

(eds.): A Study on Radio Access Technology Selection Algorithms

Wu, L., Sandrasegaran, K. (eds.): A Study on Radio Access Technology Selection Algorithms. Springer (2012)

2012
[10]

In: Advances in Vehicular Networking Technologies

Maria Vegni, A., Inzerilli, T., Cusani , R.: Seamless Connectivity Techniques in Vehicular Ad-hoc Networks. In: Advances in Vehicular Networking Technologies. InTech (2011)

2011
[11]

In: 6th IEEE International Symposium on Wireless Vehicular Communications, WiVeC 2014, Vancouver, BC, Canada, September 14-15, pp

Bansal, G., Cheng, B., Rostami, A., Sj¨oberg, K., Kenney, J.B., Gruteser, M.: Comparing LIMERIC and DCC approaches for VANET channel congestion control. In: 6th IEEE International Symposium on Wireless Vehicular Communications, WiVeC 2014, Vancouver, BC, Canada, September 14-1...

2014
[12]

IEEE Transactions on Vehicular Technology 57(4), 2426 –2439 (2008)

Hasib, A., Fapojuwo, A.: Analysis of common radio resource management scheme for end -to-end qos support in multiservice heterogeneous wireless networks. IEEE Transactions on Vehicular Technology 57(4), 2426 –2439 (2008)

2008
[13]

Vegni, A., Little, T.: Hybrid vehicular communications based on v2v-v2i protocol switching. Intl. Journal of Vehicle Information and Communication Systems (IJVICS) 2(3/4) (2011), doi:10.1504/IJVICS.2011.044263

2011 arXiv
[14]

In: Global Telecommunications Conference, GLOBECOM 2006, pp

Liu, X., Li, V., Zhang, P.: Nxg04-4: Joint radio resource management through vertical handoffs in 4g networks. In: Global Telecommunications Conference, GLOBECOM 2006, pp. 1–5. IEEE (November 2006)

2006
[15]

In: IEEE International Conference on Communications, ICC 2002, vol

Tolli, A., Hakalin, P., Holma, H.: Performance evaluation of common radio resource management (crrm). In: IEEE International Conference on Communications, ICC 2002, vol. 5, pp. 3429–3433 (2002)

2002

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

IEEE Transactions on Mobile Computing (TMC) 99 (2014)

Park, Y., Kuk, S., Kim, H., Ha, J., Liang, C.J., JeongGil, K.: A feasibility study and development framework design for realizing smartphone -based vehicular networking systems. IEEE Transactions on Mobile Computing (TMC) 99 (2014)

work page 2014

[2] [2]

EURASIP Journal on Wireless Communications and Networking 2014(1), 89 (2014)

Hameed Mir, Z., Filali, F.: Lte and ieee 802.11p for vehicular networking: A performance evaluation. EURASIP Journal on Wireless Communications and Networking 2014(1), 89 (2014)

work page 2014

[3] [3]

In: IEEE 79th Vehicular Technology Conference (VTC2014-Spring) (May 2014)

Hameed Mir, Z., Filali, F.: On the Performance Comparison between IEEE 802.11 p and LTE-based Vehicular Networks. In: IEEE 79th Vehicular Technology Conference (VTC2014-Spring) (May 2014)

work page 2014

[4] [4]

Vinel, A.: 3gpp lte versus ieee 802.11 p/wave: which technology is able to support cooperative vehicular safety applications? IEEE Wireless Communications Letters 1(2), 125–128 (2012)

work page 2012

[5] [5]

ISO/DIS 21217 (2006)

ISO: Intelligent transport systems-communications access for land mobiles (calm)architecture. ISO/DIS 21217 (2006)

work page 2006

[6] [6]

IEEE Draft Standard P802.21/D13 (2008)

IEEE: Local and metropolitan area networks: Media independent handover services. IEEE Draft Standard P802.21/D13 (2008)

work page 2008

[7] [7]

IEEE Std 802.11p-2010 (2010)

IEEE: Ieee standard for information technology –local and metropolitan area networks–specific requirements –part 11: Wireless lan (mac) and (phy) specifications amendment 6:wave. IEEE Std 802.11p-2010 (2010)

work page 2010

[8] [8]

LTE: Overview of 3gpp release 8, http://www.3gpp.org/Release -8 (2010)

work page 2010