{"id":"6a78d38b-a1c7-494b-a21f-39c15d01d418","arxiv_id":"2412.04222","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"high","formal_verification":"none","parameter_count":3,"one_line_summary":"DistB-VNET layers a dual blockchain, SDN/NFV control, and an isolation forest IDS onto VANETs, claiming 99.23% attack-detection accuracy and improved network performance in simulation.","lead":"This paper proposes a smart-city vehicle network that combines blockchain, software-defined networking, network virtualization, and an isolation-forest detector to secure traffic and manage resources. The claimed payoff is 99.23% accuracy on malicious traffic and better throughput and latency than one earlier protocol, but the supporting code and simulation details are not provided.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Table II's performance comparison is not reproducible because the baseline configuration and simulation parameters are unspecified, so the claimed improvements cannot be distinguished from artifact.","rationale":"The paper's abstract claims a 99.23% detection accuracy and that DistB-VNET 'greatly improves network performance, offering decreased latency, increased security, and reduced congestion.' The accuracy figure is presented without dataset preprocessing, hyperparameters, or code, but it is at least a plausible ML result. The network performance claim, however, is supported only by Table II and Figures 3-4, with no description of the simulation methodology. Table II compares against IEAOCGO-C but omits the baseline's configuration, the traffic and mobility models, the values of the delay parameters in Eqs. (1)-(2), and the number of simulation runs. This is exactly the kind of missing information that prevents independent verification of a comparative claim. If the baseline was run under different conditions or the numbers were taken from the original paper's different scenario, the comparison is invalid. Additionally, the unit inconsistency (ETED in mJ, ECM in ms) and the lack of any statistical dispersion measures further weaken confidence. The reader's verdict of REJECT is appropriate because the central quantitative claims cannot be validated from the manuscript. I agree with the reader's weakest assumption, and no adjustment to the verdict is needed. My concrete test would force the authors to either supply the exact simulation setup or allow an independent rerun, which would settle whether the performance improvement is real or an artifact.","tokens_in":9118,"tokens_out":3344,"duration_ms":35585,"concrete_test":"Request from the authors the complete Mininet/emulation scripts, the mobility and traffic configuration files, and the exact implementation of IEAOCGO-C used to generate Table II. If these are unavailable, independently re-implement IEAOCGO-C from reference [27] and run it under the same conditions (80 nodes, 50 m/s max speed, 100-512 byte packets, identical traffic model and duration); compare the resulting PDR, ETED, THRPT, NLT, and ECM to Table II. If the baseline's numbers change by more than 5% relative to the published values, or if DistB-VNET's relative advantage disappears, the central performance claim is unsupported.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central performance claim ('greatly improves network performance... decreased latency, increased security, and reduced congestion') rests almost entirely on Table II, which compares DistB-VNET with IEAOCGO-C across NLT, PDR, THRPT, ETED, and ECM. The manuscript does not specify the traffic model, mobility model, the delay components tb, ts, tn, tc, tv used in Eqs. (1) and (2), any blockchain gas parameters, or how the IEAOCGO-C baseline was configured and run. Section IV-B lists only emulator, node count, speed, packet size, and blockchain-related tools. Without these details, the numbers in Table II could arise from mismatched simulation scenarios, different traffic loads, or selective reporting. This is not a matter of style: the comparison is the only quantitative evidence for the headline performance improvement. The reader's concern is therefore valid. An additional strengthening issue is that Algorithm 1 lines 14-15 use encryption with the private key and decryption with the public key, which does not provide confidentiality, but the Table II reproducibility problem is the more load-bearing flaw because it undermines the primary quantitative claim of the paper.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript proposes DistB-VNET, a distributed blockchain-based vehicular ad-hoc network architecture that integrates SDN, NFV, and an unsupervised Isolation Forest intrusion detection system for smart city VANETs. The architecture uses a dual-layer blockchain (distributed for vehicle communication, centralized for cloud verification) and cluster-based communication with RSUs. The paper claims 99.23% accuracy for malicious traffic detection on the IDS 2018 dataset and, via a comparison with IEAOCGO-C in Table II, claims reduced latency, increased security, and reduced congestion. The evaluation also includes throughput equations, plots of throughput and communication overhead, and gas consumption data.","tokens_in":9363,"tokens_out":3852,"duration_ms":39011,"significance":"If the reported results are correct, the combination of edge-level unsupervised intrusion detection with blockchain-backed SDN-NFV management would be a useful contribution to secure vehicular networks. The use of a public dataset (IDS 2018) for the ML evaluation is a positive feature, and the architecture description is reasonably detailed. However, the central quantitative claims are not presently reproducible: the network performance comparison in Table II lacks simulation methodology, the ML result lacks preprocessing and hyperparameter details, and Algorithm 1 contains a cryptographic direction error. The contribution is therefore interesting in scope but not yet substantiated at the standard required for publication.","major_comments":[{"comment":"The comparison against IEAOCGO-C is the sole quantitative evidence for the claimed performance improvement, but it is not reproducible. Section IV-B lists only the emulator, blockchain platform, node count, speed, and packet size; it does not specify the traffic model, mobility model, the delay components tb, ts, tn, tc, tv used in Eqs. (1)–(2), blockchain gas parameters, or how the baseline IEAOCGO-C was configured and run. Table II also has inconsistent units: ETED is labeled mJ (millijoules, an energy unit) and ECM is labeled ms (milliseconds, a time unit). These problems mean the claimed improvements cannot be distinguished from artifact.","section":"Section IV-C, Table II"},{"comment":"The secure communication procedure performs encryption with the private key and decryption with the public key. This does not provide confidentiality; it is message signing and verification. The abstract and introduction claim confidentiality, so this is a load-bearing security error. If confidentiality is intended, the algorithm should encrypt with the public key and decrypt with the private key.","section":"Algorithm 1, lines 14–15"},{"comment":"The 99.23% accuracy on the IDS 2018 dataset is not verifiable. The paper does not report preprocessing steps, feature selection, train/test split, class imbalance handling, or Isolation Forest hyperparameters (e.g., n_estimators, max_samples, contamination), and no code or reproducible artifact is provided. Without these details, the accuracy result could reflect data leakage or an optimistic evaluation setup.","section":"Section III-A, Table IV"},{"comment":"Equations (1)–(3) are definitions with free parameters (tb, ts, tn, tc, tv, G0, Cb); they do not constitute a derivation or a measurement procedure. No values for these delays or gas coefficients are given, and the paper does not explain how they were estimated in the Mininet/Ethereum setup. Consequently, the throughput and gas-consumption claims in Figures 3–5 and Table II are not supported by these equations.","section":"Section IV-A, Eqs. (1)–(3)"},{"comment":"The text states that for each fixed cluster size, throughput decreases as the number of vehicle nodes increases. This contradicts Eq. (2), which predicts throughput increasing with V for constant D(V) and constant delay components. The paper provides no explanation of how the delay model reverses this trend, so the reported behavior is internally inconsistent with the presented throughput equation.","section":"Section IV-C, Fig. 3"}],"minor_comments":[{"comment":"There are repeated typographical issues such as 'V ANETs' with nonbreaking spaces and 'noble isolation forest' in the abstract; these should be corrected to 'VANETs' and 'novel isolation forest'.","section":"Abstract and throughout"},{"comment":"The table contains the typo 'achevied' for 'achieved', and the column formatting should be aligned for readability.","section":"Table I"},{"comment":"The caption says 'anf f-score' and should read 'and F1-score'.","section":"Table IV"},{"comment":"The 'Max. vehicle speed' parameter is listed as 50 m/s^2; the unit should be m/s, not an acceleration unit.","section":"Section IV-B, Table III"},{"comment":"The IDS 2018 dataset reference should specify the exact subset of the CSE-CIC-IDS2018 data used, the access date, and any filtering or normalization applied.","section":"Reference [23]"}],"recommendation":"major_revision","confidential_remarks":"This manuscript needs substantial revision. The missing simulation methodology for Table II and the cryptographic direction error in Algorithm 1 are the most serious issues. If the authors cannot provide the missing experimental details and correct the security flaw, the paper should not be published."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis is a paper about composing blockchain, SDN, NFV, and isolation forest into a VANET architecture for smart cities. If you have read one blockchain-SDN-VANET survey, you have seen every component here. The only new bit is the \"dual-layer\" blockchain (one distributed among vehicles, one centralized in cloud), but that's a design choice, not a mechanism.\n\nWhat the paper does well: The architecture is described clearly enough to follow, and the authors are honest up front that the system has not yet been tested under higher transmission costs and gas fees. The accuracy number (99.23% for isolation forest) is entirely plausible; it's the kind of score one gets on the IDS 2018 dataset, and the paper correctly notes the model is unsupervised.\n\nNow the soft spots, in order of severity.\n\nFirst, the headline performance claim — decreased latency, increased throughput, reduced congestion — rests on Table II comparing the proposed system with IEAOCGO-C across five metrics. The manuscript never says how the baseline was configured, what traffic or mobility model was used, what delay values (tb, ts, tn, tc, tv) went into Eqs. (1)-(2), or what blockchain gas parameters were set. With that little information, the numbers could be artifacts of different simulation loads or selective reporting. That is the load-bearing flaw.\n\nSecond, the equations themselves are definitions, not derivations: throughput as data divided by delay sum, gas as a linear function of transactions. They don't predict anything. Third, the pseudocode in Algorithm 1 encrypts with the private key and decrypts with the public key, which gives authentication, not confidentiality — and confidentiality is the claimed property. That is a genuine error, though it may be a typo. Finally, the table units are inconsistent (ETED in mJ, ECM in ms), and no code, data, or hyperparameters are released for the ML result.\n\nThe citation pattern includes many self-citations to the authors' earlier blockchain-SDN work. That is not itself a flaw, since those papers are related, but it does not add support here.\n\nWho is this for? A reader wanting a quick conceptual sketch of how these technologies could be combined. But the paper lacks the reproducible evidence needed to be taken seriously as a research result. I would not send it to peer review; it needs substantial additional work: a real simulation description, baseline configuration, and a corrected algorithm — and even then the novelty is thin.\n\nHappy to discuss over coffee.","headline":"A plausible architecture sketch with an unreproducible performance table and a cryptographic error; not referee-ready.","tokens_in":9908,"tokens_out":2303,"would_cite":false,"duration_ms":22190,"reading_group":"no","serious_thinker":"no","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A blockchain-SDN-NFV VANET architecture claims 99.23% accuracy in blocking malicious traffic while improving latency, throughput, and congestion in smart-city vehicle networks.","keywords":["blockchain","software-defined networking","network function virtualization","vehicular ad-hoc networks","intrusion detection","isolation forest","smart city","vehicle-to-everything"],"falsifier":"Re-running the simulation with a stated traffic model, explicit delay values for blockchain, SDN, NFV, and intra-cluster communication, and the same configuration for both systems: if the proposed system fails to beat IEAOCGO-C on at least one of the reported metrics, or if the 99.23% isolation-forest accuracy drops when evaluated on a held-out subset of IDS 2018, the central claims fail.","tokens_in":8933,"feed_emoji":"🚗","tokens_out":1821,"duration_ms":19341,"temperature":0.7,"pith_summary":"The paper argues that combining a dual-layer blockchain, SDN-NFV management, and an unsupervised isolation-forest intrusion detector yields a vehicular ad-hoc network that is both secure and efficient for smart cities. The core promise is that this integration detects and blocks malicious traffic with 99.23% accuracy, while also reducing end-to-end delay and energy consumption and increasing throughput and packet delivery compared with an existing optimization protocol. If correct, it would provide a practical template for secure, scalable vehicle-to-infrastructure communication under real urban traffic loads.","feed_headline":"Blockchain VANET claims 99.23% accuracy in blocking malicious traffic","feed_subtitle":"A dual-layer blockchain plus SDN-NFV design also reports lower latency and higher throughput in smart-city vehicle simulations.","key_machinery":"The central mechanism is a three-stage pipeline: an unsupervised isolation-forest classifier placed after the cluster head filters edge traffic and blocks malicious flows; a dual-layer blockchain (distributed among vehicles, centralized in the cloud) ensures tamper-resistant data integrity and two-step verification; and SDN-NFV controllers dynamically allocate resources and manage routing through cluster heads. Isolation forest is a tree-based anomaly detector that isolates outliers rather than profiling normal behavior, which the paper justifies as necessary for detecting novel attacks that supervised models miss.","core_discovery":"The authors claim that their DistB-VNET architecture—a cluster-based VANET where a distributed blockchain secures vehicle-to-vehicle communication, a centralized cloud blockchain verifies and stores data, SDN separates control from data planes, and NFV automates resource allocation—achieves better security and network performance than current alternatives. The isolation-forest model, trained on the IDS 2018 intrusion dataset in an unsupervised manner, is reported to reach 99.23% accuracy, 99.14% precision, 99.15% recall, and 99.07% F1-score for classifying benign versus malicious traffic. Against the IEAOCGO-C baseline, the proposed system reports higher network lifetime, packet delivery ratio, and throughput, with lower end-to-end delay and energy consumption across 20–50 vehicles.","pith_inferences":["A natural testable extension is to evaluate the same isolation-forest placement on current benchmark datasets such as CICIDS2017 or 2019, which include more recent attack families, to see whether the 99.23% accuracy persists outside the IDS 2018 distribution.","The architecture's separation of control and data planes combined with edge anomaly detection could generalize beyond vehicles to other IoT domains, such as smart-grid or healthcare monitoring, where low latency and novel-attack detection are both critical.","The paper's own conclusion admits the system has not been tested under rising transmission costs and gas fees as vehicle numbers grow, so the scalability claim remains conditional on those costs not eroding the throughput advantage."],"forward_implications":["Smart-city vehicle networks could use this architecture to block previously unseen malicious traffic at the edge before it reaches the core network.","If the performance gains are reproducible, combining blockchain with SDN-NFV could become a standard design pattern for secure vehicular communication, pushing toward lower-latency traffic management.","The reported throughput equations offer a simple quantitative model for how blockchain, SDN, and NFV delays jointly affect vehicular throughput as cluster size or vehicle count grows.","The unsupervised detection approach implies that the IDS can adapt to evolving cyber-attacks without requiring frequent retraining on labeled attack data."],"supporting_citations":[{"why":"The IDS 2018 intrusion dataset is the training and evaluation data for the isolation-forest model that produces the 99.23% accuracy claim.","marker":"[23]"},{"why":"The IEAOCGO-C algorithm is the baseline against which DistB-VNET's throughput, packet delivery, delay, and energy metrics are compared in Table II.","marker":"[27]"},{"why":"The Mininet network emulator and Ethereum blockchain environment described in this reference constitute the simulation setup used to evaluate the proposed architecture.","marker":"[28]"},{"why":"The earlier DistB-Condo distributed blockchain-SDN model is the architectural foundation that DistB-VNET extends with NFV and cluster-based vehicular operation.","marker":"[25]"},{"why":"This reference supports the design of blocking malicious traffic and temporarily storing it until cleared, a mechanism the proposed verification process relies on.","marker":"[24]"}],"fun_headline_variants":["Blockchain VANET scores 99.23% in malicious traffic test","Dual-layer blockchain VANET: 99.23% accuracy, lower latency","SDN-NFV + blockchain VANET improves reliability and speed","Isolation forest IDS in VANET detects 99.23% malicious traffic","DistB-VNET: Blockchain VANET achieves 99.23% detection"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central performance comparison assumes that the simulation produced realistic, comparable numbers for both DistB-VNET and the IEAOCGO-C baseline, even though the paper does not specify the traffic model, blockchain gas parameters, or how the baseline was configured.","fun_headline_variants_meta":{"raw":{"variants":["Blockchain VANET scores 99.23% in malicious traffic test","Dual-layer blockchain VANET: 99.23% accuracy, lower latency","SDN-NFV + blockchain VANET improves reliability and speed","Isolation forest IDS in VANET detects 99.23% malicious traffic","DistB-VNET: Blockchain VANET achieves 99.23% detection"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001141,"raw_usage":{"total_tokens":4734,"prompt_tokens":941,"completion_tokens":3793,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":557,"completion_tokens_details":{"reasoning_tokens":3694}},"tokens_in":557,"tokens_out":3793,"duration_ms":28492,"temperature":1.0,"reasoning_tokens":3694,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T21:37:29.560724+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-running the simulation with a stated traffic model, explicit delay values for blockchain, SDN, NFV, and intra-cluster communication, and the same configuration for both systems: if the proposed system fails to beat IEAOCGO-C on at least one of the reported metrics, or if the 99.23% isolation-forest accuracy drops when evaluated on a held-out subset of IDS 2018, the central claims fail.","supporting_citations":[{"cited_title":"Ids intrusion csv dataset,","cited_arxiv_id":null,"evidence_quote":"The IDS 2018 intrusion dataset is the training and evaluation data for the isolation-forest model that produces the 99.23% accuracy claim."},{"cited_title":"Intelligent energy aware optimization protocol for vehicular adhoc networks,","cited_arxiv_id":null,"evidence_quote":"The IEAOCGO-C algorithm is the baseline against which DistB-VNET's throughput, packet delivery, delay, and energy metrics are compared in Table II."}],"review_version":1}