{"id":"11ebad29-5755-4cfa-ade5-dd4f89a4a94f","arxiv_id":"2411.14539","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A MATLAB simulation of a two-row multi-hop wireless network shows that network-coded transmission achieves 50% to 100% higher Shannon capacity than traditional relaying across one or two streams.","lead":"This paper simulates a simple two-row multi-hop wireless network and compares traditional store-and-forward relaying with network-coded transmission under interference. It reports that network coding raises modeled throughput by 50% to 100% while reducing the impact of extra hops, and that choosing the right transmission schedule matters.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"NC capacity formula sums rates of a single XOR broadcast; should be 2·min(R_f,R_r)/Z, and the 2/Z steady-state rate is not proven for all N_o,Z.","rationale":"The reader correctly targeted the /Z versus /2Z normalization, but the deeper issue is that the numerator is also wrong: a single XOR broadcast cannot simultaneously carry two independent rates R_f and R_r; it is limited by the weaker of the two links. The paper's model uses Shannon capacity per link independently, which is only valid for unicast transmissions, not for a broadcast that must be decodable by both receivers. This is a concrete mathematical error, not merely a question of protocol overhead. The qualitative conclusion that NC helps may survive a correction, but the quantitative 50–100% figure and the reported optimal Z values could shift. The paper provides no code or data, so the correction cannot be checked post hoc; the proposed simulation would settle both the sum-vs-min issue and the general achievability of 2/Z. I do not escalate to REJECT because the model is simple and the flaw is addressable, but the manuscript must be revised to justify or replace the capacity formula before the numerical claims are accepted. The reader's CONDITIONAL verdict remains appropriate, so no verdict change is recommended.","tokens_in":12027,"tokens_out":12104,"duration_ms":121400,"concrete_test":"Implement the exact schedule and XOR protocol of §III.C–D in a discrete-event simulator for all N_o=3,4,5,6 and Z=2..5, with link rates set from the paper's SINR model, and measure steady-state delivered packets per Z slots. Then compare Table 4 against both the published formula (R_f+R_r)/Z and the corrected formula 2·min(R_f,R_r)/Z. If measured throughput matches the corrected formula (or falls below 2/Z), the reported NC capacities and the 50–100% gain are overstated.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing step is the capacity normalization in §III.A: NC capacity = (R_f+R_r)/Z and TR = (R_f+R_r)/(2Z). The factor-of-two NC advantage depends on two unproven assumptions. First, the paper asserts without general derivation that in steady state a relay can always XOR a newly received packet with a stored one, so each direction sustains one packet per Z timeslots for every N_o∈{3..6}, Z∈{2..5}; only one hand-run example (N_o=5, Z=4, §III.D) is given, and no packet-level simulation verifies this rate for the other configurations. Second, even if one packet per Z is sustained, the numerator R_f+R_r is not the correct capacity for the described XOR broadcast: in a slot where a relay transmits A⊕B to both neighbours, both receivers must decode the same packet at the same rate, so the transmission is limited by min(R_f,R_r), and the total useful throughput is 2·min(R_f,R_r)/Z, not (R_f+R_r)/Z. When the two directions have different bottleneck SINRs — which the paper's own asymmetric latencies (4 vs 10 slots in Fig. 7) show can happen — the sum formula overestimates NC capacity. The claimed 50–100% margin therefore is not established by the model as written.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents a simplified analytical model, implemented in MATLAB, that compares traditional relaying (TR) and network coding (NC) in infrastructure-less multi-hop wireless networks. The network layout consists of one or two parallel rows of equally spaced nodes, and the model computes per-link SINR using an inverse-power path-loss law, maps SINR to capacity via the Shannon formula, and aggregates per-direction rates over a scheduling period to obtain a network capacity figure. The central claim is that NC achieves 50% to 100% higher Shannon capacity than TR in all simulated configurations (hop counts 2 to 5, scheduling periods 2 to 5, one or two streams). The paper also reports that increasing hop count reduces capacity and that two coexisting streams degrade performance relative to a single stream.","tokens_in":12242,"tokens_out":12012,"duration_ms":110372,"significance":"The paper's main strength is that it provides a concrete, parameterized calculation chain from physical-layer interference to end-to-end capacity in a simple multi-hop setting, and it explicitly incorporates transmission scheduling and interference, which many qualitative treatments of network coding omit. If the capacity aggregation and steady-state scheduling assumptions were correct, the quantitative claim of 50–100% improvement would be a useful baseline for tactical ad-hoc network design. However, the load-bearing capacity formula for NC is not correct as written, and the NC scheduling algorithm is not defined or verified for the full parameter sweep. The paper also does not provide code, a complete schedule specification, or packet-level verification, which limits reproducibility. The qualitative direction of the result (NC can outperform TR under interference) is plausible, but the specific quantitative claims in the abstract and Section VI are not established by the current model.","major_comments":[{"comment":"The capacity normalization for NC is not justified. The text states that the NC capacity per timeslot is (R_f + R_r)/Z, whereas the described NC operation has a relay transmit a single XOR-coded packet to both adjacent receivers in the same timeslot, as in Scenario 2 (§III.D, TS4). A single broadcast transmission cannot support two different rates toward the two receivers; the achievable common rate is min(R_f, R_r). The total useful throughput in steady state is therefore at most 2·min(R_f, R_r)/Z, not (R_f + R_r)/Z. The two directions can have substantially different SINR because the interference environment differs between forward and reverse transmissions, as reflected in the different latencies in Eqs. (13) and (14). The reported 50–100% gains in Table 4 are thus not established by the model as written. The authors should either correct the NC capacity formula, or justify that the two directions are scheduled in separate timeslots (which contradicts the described XOR broadcast), or provide a worst-case analysis using min(R_f, R_r).","section":"III.A"},{"comment":"The steady-state assumption that NC sustains one packet per direction per Z timeslots for all N_o and Z is only illustrated by the hand-run example N_o = 5, Z = 4 in Scenario 2 and Figure 7. No general proof or packet-level simulation is provided for the other configurations used in the results (N_o = 3..6, Z = 2..5). Moreover, the NC schedule is not formally defined in the way TR is defined by Eqs. (7) and (8). The example interleaves reverse-direction transmissions (e.g., node 2 transmitting to node 1 in TS10 of Figure 7) in a manner that is not captured by the 'simple sequential forward schedule' mentioned in §III.C. Without a precise schedule for each timeslot and a proof of the one-packet-per-direction-per-Z-timeslot property, the capacity results for the entire parameter sweep in Table 4 are unsupported. The authors should provide a general scheduling algorithm for NC and verify the steady-state rate for all simulated N_o and Z values, either analytically or by simulation.","section":"III.C and III.D"}],"minor_comments":[{"comment":"The Introduction promises a Section IV on the process for computing SINR, but the manuscript jumps from Section III.D directly to Section V. The SINR computation is only referred to in passing and via Eq. (6), which hampers reproducibility.","section":"General structure"},{"comment":"Figures 6 and 7 are each used twice: once in Section III.D (scheduling scenarios) and again in Section V (capacity plots). This creates confusion and should be corrected by renumbering.","section":"Figures"},{"comment":"The values in the last column of Table 4 have inconsistent signs. For example, for N = 3, the TR row shows -15% but the numbers imply (C_OS - C_TS)/C_TS = +18%. The authors should check the sign convention and the associated formulas for this column.","section":"Table 4"},{"comment":"The reference list jumps from [18] to [21]; entries [19] and [20] are missing. In addition, some cited works date from 2005–2015 and more recent literature on network coding and scheduling could be cited.","section":"References"},{"comment":"Equation numbering skips: Eq. (2) and Eq. (5) are not present in the text, though Eq. (1), (3), (4), and (6) are referenced. This should be corrected for clarity.","section":"Equations"},{"comment":"There is a typo in Scenario 2: 'node 4 transmits the NC packet A⊕B to both node and node 5' should read 'to both node 3 and node 5'.","section":"Section III.D"},{"comment":"The claim that NC outperforms TR 'under the same network settings' is imprecise, because Table 4 compares each scheme at its own optimal scheduling period Z. The comparison of optima is a legitimate approach, but it should be stated explicitly so that the reader does not assume identical Z.","section":"Abstract and Section VI"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is rough in its current form: a missing section, duplicate figure numbers, inconsistent table signs, and missing references will all need attention. More substantively, the NC capacity formula is not correct for a single XOR broadcast, and the scheduling algorithm is not fully specified. These are fixable within the scope of the paper, so I am not recommending rejection, but the authors should be asked to correct the model and either prove or simulate the steady-state schedule for all reported configurations. The paper's fit with the journal's networking scope is acceptable, though the novelty is modest given that the core network-coding concept is well established."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague, the short version: this is a textbook result (network coding beats plain relaying) wrapped in a simple deterministic simulation, and the paper's central quantitative claim rests on a capacity formula that overstates the NC gain. The qualitative finding is not new; the authors themselves cite Liew and Zhang for the two-way exchange.\n\nWhat the paper does well: it lays out a clean two-row topology, computes SINR with standard path-loss and Shannon formulas, and sweeps the scheduling period for one and two streams. The reported percentages in Table 4 are internally consistent with the equations as written. That is a real, if modest, piece of engineering reconnaissance.\n\nThe soft spots, in order of severity. First, the load-bearing normalization in Section III.A: NC capacity is (R_f+R_r)/Z, TR is (R_f+R_r)/(2Z). The factor of two encodes the scheduling advantage, but the NC numerator is wrong for an XOR broadcast. A relay sends one coded packet to two neighbors; both must decode it at the same rate, so the useful throughput is 2·min(R_f,R_r)/Z, not (R_f+R_r)/Z. When the two directions have different bottleneck rates — which the paper's own asymmetric latencies in Fig. 7 suggest is possible — the sum overestimates NC capacity. The claimed 50–100% margin is therefore not established by the model as written. Second, the steady-state assumption that one packet per direction arrives every Z slots is asserted for all N_o and Z, but only one hand-worked example (N_o=5, Z=4) is given; no packet-level simulation or proof covers the other configurations. Third, the manuscript is structurally messy: Section IV is missing, several figures are duplicated, the reference list skips numbers, Table 4's columns are hard to map, and the conclusion calls simulation \"experimental evidence,\" which it is not.\n\nNone of this is fatal to the qualitative conclusion, but it means the quantitative headline should not be taken at face value. The good news is the flaws are fixable: correct the NC capacity formula, rerun the sweeps, and if the margin survives you have a modest engineering note. If it doesn't, the abstract needs rewriting.\n\nWho is this for? A reader who wants a quick parameter-sweep baseline for scheduling period in a two-row relay network. I would not cite it, and I wouldn't bring it to reading group. As submitted, I wouldn't send it to peer review; I'd invite a resubmission after the capacity formula is fixed.","headline":"A simple simulation study with a textbook claim, undercut by a capacity-formula error that likely inflates the network-coding advantage.","tokens_in":12793,"tokens_out":6233,"would_cite":false,"duration_ms":58773,"reading_group":"no","serious_thinker":"yes","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Network-coded relays give 50–100% higher capacity than store-and-forward in modeled infrastructure-less multihop networks.","keywords":["network coding","infrastructure-less wireless networks","multi-hop relays","transmission scheduling","capacity bound","SINR","XOR network coding","throughput comparison"],"falsifier":"Re-run the same two-row topology as a packet-level experiment: measure the steady-state number of packets delivered per direction per schedule period with and without small XOR headers and finite relay buffers. If network-coded transmission cannot sustain one packet per direction every $Z$ timeslots, or if its measured throughput gain over traditional relaying falls below the model's 50% for a four-hop single-stream network, the zero-overhead steady-state assumption is contradicted.","tokens_in":11766,"feed_emoji":"📡","tokens_out":11298,"duration_ms":96270,"temperature":0.7,"pith_summary":"The paper sets out to show that, in an infrastructure-less multi-hop wireless network, relaying data with bitwise-XOR network coding yields a substantially higher capacity bound than traditional store-and-forward relaying under identical interference conditions. Its simulations cover a deliberately simple two-row topology with one or two simultaneous communication streams and two to five hops, with a periodic transmission schedule of period $Z$ controlling which nodes may transmit. Across every tested configuration, network-coded transmission delivered the higher throughput, with gains of 50% to 100% for a single stream and 52% to 88% for two coexisting streams. If the result holds in practice, network coding plus scheduling is a low-complexity way to push more data through interference-limited tactical and disaster-response networks without changing topology or transmit power.","feed_headline":"XOR network coding beats store-and-forward by 50-100%","feed_subtitle":"XOR relaying delivers one packet per direction per schedule period, roughly twice the store-and-forward rate.","key_machinery":"The load-bearing mechanism is the combination of a periodic schedule of period $Z$ with relay-side bitwise XOR combining. The schedule limits simultaneous transmitters so that in timeslot $i$, nodes $i, i+Z, i+2Z, \\dots$ transmit, which controls physical interference. In traditional mode, forward and reverse schedules alternate and the stream delivers one packet in each direction every $2Z$ timeslots; in network-coded mode, a relay XORs a newly received packet with a stored one, so one packet can travel left and one right in the same period, delivering two packets every $Z$ timeslots. The XOR algebra is just the identity $A \\oplus B \\oplus B = A$, which lets each source recover the other's packet from the combined transmission. This mechanism converts a scheduling choice into a throughput difference: fewer simultaneous transmitters raise SINR, while network coding converts the schedule into a doubled delivery rate per timeslot.","core_discovery":"The central discovery is a capacity-doubling scheduling effect. In traditional relayed transmission, forward and reverse traffic alternate under the half-duplex constraint, so two packets, one in each direction, require $2Z$ timeslots and the stream capacity is $(R_f+R_r)/(2Z)$. With network coding, relay nodes transmit a bitwise XOR of a packet from the left and a packet from the right, so both directions advance in the same schedule period and two packets arrive every $Z$ timeslots, giving capacity $(R_f+R_r)/Z$. The model computes each node's SINR from all simultaneous transmitters, converts it to a per-node rate with the capacity formula, and takes the minimum rate along the route in each direction. In all simulations the network-coded schedule produced higher capacity than traditional relaying, and the gap was largest for long routes; the advantage also persisted when a second stream added cross-stream interference, with smaller percentage degradation than traditional relaying.","pith_inferences":["Editorial inference: the modeled factor-of-two timeslot advantage is an upper bound; real XOR operations, headers, and relay buffering overhead would erode the 50–100% gain, so the numbers should be read as best-case for a clean protocol.","Editorial inference: the mechanism is not tied to the two-row layout; any topology where relays can pair left- and right-moving packets should show a similar scheduling advantage, provided SINR is computed from simultaneous transmitters.","Editorial inference: a direct extension would be to add a third parallel stream; the paper's observed pattern that network coding degrades less than traditional relaying with added interference suggests the relative advantage could persist, but additional interference rows could also change the optimum schedule period."],"forward_implications":["For a single communication stream with two to five hops, the simulated network-coded capacity exceeds traditional relaying by 50% to 100% at the respective optimum scheduling periods.","For two coexisting streams, the network-coded advantage remains 52% to 88%, showing that network coding is less degraded by cross-stream interference than traditional relaying.","The optimum scheduling period is topology-dependent: traditional relaying favors $Z=2$ at two hops and moves to $Z=3$ or $Z=4$ on longer routes, while network coding favors $Z=4$ for one stream and $Z=3$ for two streams in the tested range.","Increasing the number of hops lowers the attainable per-node data rate under both schemes, yet the relative gain of network coding persists across hop counts."],"supporting_citations":[{"why":"Supplies the XOR-exchange principle used to show one relay transmission can serve both directions.","marker":"[11]"},{"why":"Establishes that channel capacity drops with carrier-to-interference ratio, the basis for computing throughput from SINR.","marker":"[3]"},{"why":"Provides the physical-interference scheduling concept the paper adapts to define which nodes transmit in each timeslot.","marker":"[22]"},{"why":"Supplies the bitwise XOR packet-combining method used by network-coded relays.","marker":"[23]"},{"why":"Gives the inverse-power-law path loss equation used to compute received signal and interference power.","marker":"[12]"},{"why":"Motivates the half-duplex constraint behind the minimum scheduling period $Z=2$.","marker":"[13]"}],"fun_headline_variants":["Network coding doubles multi-hop wireless throughput","XOR relaying outshines store-and-forward in mesh nets","Coding beats forwarding: double the capacity in ad-hoc nets","Network-coded mesh: higher capacity, less interference hit","XOR trick boosts multi-hop wireless capacity by ~2x"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The network-coding advantage rests on assuming a relay always has a stored packet to XOR with the newly received packet every schedule period, with no protocol overhead; if a relay must wait or send extra headers, the factor-of-two rate gain shrinks.","fun_headline_variants_meta":{"raw":{"variants":["Network coding doubles multi-hop wireless throughput","XOR relaying outshines store-and-forward in mesh nets","Coding beats forwarding: double the capacity in ad-hoc nets","Network-coded mesh: higher capacity, less interference hit","XOR trick boosts multi-hop wireless capacity by ~2x"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000242,"raw_usage":{"total_tokens":1500,"prompt_tokens":892,"completion_tokens":608,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":508,"completion_tokens_details":{"reasoning_tokens":527}},"tokens_in":508,"tokens_out":608,"duration_ms":5765,"temperature":1.0,"reasoning_tokens":527,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:09:23.824355+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Re-run the same two-row topology as a packet-level experiment: measure the steady-state number of packets delivered per direction per schedule period with and without small XOR headers and finite relay buffers. If network-coded transmission cannot sustain one packet per direction every $Z$ timeslots, or if its measured throughput gain over traditional relaying falls below the model's 50% for a four-hop single-stream network, the zero-overhead steady-state assumption is contradicted.","supporting_citations":[{"cited_title":"Physical-layer Network Coding: Tutorial, survey, and beyond","cited_arxiv_id":null,"evidence_quote":"Supplies the XOR-exchange principle used to show one relay transmission can serve both directions."},{"cited_title":"Interference in Wireless Multi-Hop Ad-Hoc Networks and Its Effect on Network Capacity","cited_arxiv_id":null,"evidence_quote":"Establishes that channel capacity drops with carrier-to-interference ratio, the basis for computing throughput from SINR."},{"cited_title":"Wireless link scheduling under physical interference model,","cited_arxiv_id":null,"evidence_quote":"Provides the physical-interference scheduling concept the paper adapts to define which nodes transmit in each timeslot."},{"cited_title":"Network Coding for Wireless Applications: A review","cited_arxiv_id":null,"evidence_quote":"Supplies the bitwise XOR packet-combining method used by network-coded relays."},{"cited_title":"Wireless Communications","cited_arxiv_id":null,"evidence_quote":"Gives the inverse-power-law path loss equation used to compute received signal and interference power."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Motivates the half-duplex constraint behind the minimum scheduling period $Z=2$."}],"review_version":1}