{"id":"29a82860-a046-4079-a618-0a316bf2504d","arxiv_id":"2412.03956","paper_version":5,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For bistatic ISAC with dispersed receivers and a sensor that does not know the communication messages, BIA and TIM schemes achieve (sDoF,cDoF) tradeoff points that dominate time-sharing.","lead":"This paper applies two known interference management techniques, blind interference alignment and topological interference management, to bistatic integrated sensing and communication systems. It derives new sensing-versus-communication tradeoff points that beat simple time-sharing between sensing-only and communication-only operation.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1's BIA cancellation requires the sensing channel to be exactly constant within each block, while the model only promises 'approximately constant'; a fixed drift leaves O(P) residual communication interference and destroys the claimed sDoF.","rationale":"The reader's weakest assumption matches the most load-bearing concern: the entire BIA sensing gain hinges on exact constancy of the sensing channel within a block, while the paper's model text only promises approximate constancy. The proof uses exact equality, and no quantitative robustness is provided. This is a genuine gap between the physical justification and the mathematical requirement: a small fixed drift creates interference that scales with the communication power, so at high SNR the sensing task fails to achieve the stated sDoF. The rest of the Theorem 1 construction is internally consistent under the exact-constancy assumption: the receiver ZF step uses K i.i.d. channel vectors and is generically full rank; the sensor differencing step uses the designed x_s structure to isolate known projections of the sensing channel; and the cDoF count of 1 is correctly obtained by sending two symbols per transmitter over 2K slots. The proof typos in Theorem 4 and the MU-MIMO appendix are real but fixable and are secondary to the central Theorem 1 claim. The reader's CONDITIONAL verdict remains appropriate: the achievability proof is plausible under ideal assumptions, but the sensitivity of the central cancellation to the stated 'approximately constant' assumption, together with the unquantified drift and the formal inconsistencies noted by the reader, means the paper should not be accepted without revision.","tokens_in":33013,"tokens_out":28253,"duration_ms":283609,"concrete_test":"Perturb the model in Theorem 1 by setting h_s(t) = h + e(t) with sup_t ||e(t) − e(1)|| ≤ ε for a fixed ε > 0, while keeping the communication symbols at power P. Derive the post-cancellation sensor signal y_s(1) − y_s(t); show that the communication-induced residual has variance Θ(ε^2 P). Compute the asymptotic channel-estimation error as P → ∞ for fixed ε > 0. If the error does not vanish, the claimed sDoF point is not robust to any fixed sensing-channel drift, confirming that the proof requires exact equality and not merely approximate constancy.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central construction in Section IV-A cancels the sensor's communication interference by subtracting received signals, e.g., y_s(1) − y_s(t). This cancellation is exact only if h_{si}(1) = h_{si}(t) for every sensing-channel coefficient and every t in the block, and if the repeated communication symbol is exactly the same in those slots. Section II-A.1, however, states that the sensor channel 'remains approximately constant' over t0, and the proof uses exact equality (e.g., 'Since h[K+1](1) = ... = h[K+1](a)'). If the sensing channel actually drifts, writing h(t) = h + δ(t) with fixed nonzero drift δ, the subtraction leaves a residual Σ_i δ_i(t) W^i whose power scales as P, because the unknown communication symbols W^i have power scaling with P. The residual therefore does not vanish at high SNR, so the sensor observes h x^0 + noise plus a nonvanishing interference term; the claimed sDoF = (K−1)/K is not obtained. The paper supplies no bound on the drift rate that would make the residual negligible, and no robustness analysis. This is load-bearing because the advertised improvement over time-sharing rests entirely on this cancellation, and it is exactly the point at which the physical 'approximately constant' assumption meets the mathematical 'exactly constant' requirement.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies a bistatic integrated sensing and communication (ISAC) system with multiple transmitters, dispersed communication receivers, and one single-antenna sensor. The communication messages are unknown to the sensor and act as interference, while the transmitters do not know the sensor channel. The paper proposes two interference-management strategies: blind interference alignment (BIA) for channels with heterogeneous coherence times (interference channel, MU-MISO, MU-MIMO) and topological interference management (TIM) for heterogeneous connectivity ((K+1,U,D) neighboring antidotes and (K+1,d) regular networks). The main results are achievable tradeoff points between the sensing degree of freedom (sDoF) and the communication degree of freedom (cDoF), for example ((K-1)/K, 1) for the K-user interference channel with a sensor, which improve on the time-sharing line between sensing-only and communication-only points. The paper also provides simulation results comparing channel estimation error with TIN and SIC benchmarks.","tokens_in":33299,"tokens_out":38190,"duration_ms":381568,"significance":"If the results hold, the paper makes a useful contribution by showing that bistatic ISAC can exploit coherence-time or connectivity heterogeneity to cancel communication interference at the sensor without sensor CSI, thereby adding sensing capability at no loss of communication DoF. The BIA construction for the K-user interference channel is elegant, and the TIM extension to ISAC is natural. The paper is largely self-contained, with explicit linear-algebra achievability proofs and a link to simulation code. However, the formal state of the proofs, especially for Theorem 4, and the gap between the stated 'approximately constant' sensing-channel assumption and the exact-constancy requirement in the BIA cancellation steps prevent verification of the claims in the current form.","major_comments":[{"comment":"The model in Section II-A.1 states that the sensing channel 'remains approximately constant' over t0, but the BIA cancellation step (e.g., the subtraction after Eq. (11) in Section IV-A, and the equality h[K+1](1)=...=h[K+1](a) in Section IV-B) requires exact constancy of every sensing-channel coefficient within each block. With a fixed drift h(t)=h+δ(t), the difference y_s(1)-y_s(t) leaves the residual sum over i of δ_i(t) W^i, whose power scales as P because the communication symbols W^i have power P; the claimed sDoF=(K-1)/K in Theorem 1, and the analogous claims in Theorems 2 and 3, are then not achieved. The paper supplies no drift-rate condition under which the residual becomes negligible at high SNR. Please either state the model as exact block-constant sensing channels or provide a robustness analysis with an explicit drift bound.","section":"II-A.1, IV-A, IV-B"},{"comment":"The proof of Theorem 4 is internally inconsistent. It sets t0=(K-D-U+1)⌈(K-D+U+1)/(U+1)⌉ and N=(U+1)⌈(K-D+U+1)/(U+1)⌉, which give sDoF=N/t0=(U+1)/(K-D-U+1), not the claimed (U+1)/(K-D+U+1) in Eq. (5). The sensing signals x[0]_1,...,x[0]_N are defined in C^{(K-D-U+1)×1} but the orthogonality condition is written with I_{(K-D+U+1)}. In addition, the text says 'U linearly independent left null vectors' but then uses U+1 vectors v0,1,...,v0,U+1 and claims U+1 effective observations. The example (K=5,U=1,D=2) has t0=10 and N=4, which is consistent with the corrected formulas t0=(K-D+U+1)⌈(K-D-U+1)/(U+1)⌉ and N=(U+1)⌈(K-D-U+1)/(U+1)⌉. These corrections must be made and the proof re-verified.","section":"V-A (Theorem 4)"},{"comment":"The proof of Theorem 5 states that the sensing signals x[0]_i lie in C^{(d+1)×1} and satisfy orthogonality with I_{d+1}, but the construction x[0]_i=[v0,i x^{[K+1]}_s, v0,i x^{[1]}_s, ..., v0,i x^{[d-1]}_s]^T has only d components, since the sensor is connected to d transmitters. The dimension should be d and the identity should be I_d. Furthermore, the claim that receivers outside the set {d,...,K-d+1} can also decode 2 messages because their interference dimension 'does not exceed d-1' is asserted without a calculation; a short derivation for these boundary receivers is needed to complete the proof.","section":"V-B (Theorem 5)"}],"minor_comments":[{"comment":"In the sentence 'we let each receiver decode U+1 communication message symbols, and let the sensor obtain U effective observations', the final 'U' should be 'U+1' to agree with the rest of the proof and with Example 2.","section":"V-A"},{"comment":"There are numerous typos, including 'Hall = ... ∈ C^{m×1}' (should be C^{m×m}), 'anttenas', 'siganl', 'vertors', and 'symblos'. The stacked matrix Hall is described confusingly; please clarify its dimensions and the notation H[S+1,1](b)[1:q,:].","section":"Appendix A (Theorem 3)"},{"comment":"There is a stray punctuation sequence '; ,' at the end of the theorem statement; remove it.","section":"III-B, Theorem 4"},{"comment":"In the channel output equations, the connectivity parameter g_ki multiplies the noise term as well as the signal; noise should be present regardless of connectivity, so g_ki should multiply only the signal term.","section":"II-A"},{"comment":"The sentence 'by obtaining 4 effective observations can be obtained over 6 time slots' is ungrammatical; please rewrite.","section":"IV-A, Example 1"},{"comment":"The phrase 'the achieved ISAC tradeoff points ... are characterized' is stronger than what is proved, since the paper provides achievability without a converse. Suggest rewording to 'achievable tradeoff points are provided'.","section":"Abstract and Section I"},{"comment":"The phrase 'the transmitters emits' should be 'the transmitters emit'.","section":"II-B.2"},{"comment":"The orthogonality condition for the sensing signals is written with a bare 'I' without subscript; please specify the correct identity size.","section":"Appendix B-A"}],"recommendation":"major_revision","confidential_remarks":"The main idea is promising and the examples suggest that the intended results are correct, but the formal state of the TIM proofs and the model-assumption gap in the BIA proofs require substantive revision. I do not see grounds for rejection, provided the authors can fix the algebra and either tighten the model assumption or provide robustness guarantees. Please also check the boundary case K=U+D in Theorem 4, where the set Rs notation degenerates and the theorem's communication-only point may be suboptimal relative to the known results for that parameter regime."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper is a genuine extension, not a re-derivation: it brings blind interference alignment and topological interference management to bistatic ISAC where the sensor does not know the communication messages or the sensing-channel CSI, and it produces closed-form (sDoF,cDoF) tradeoff points that beat time-sharing. The cleanest is Theorem 1, the K-user interference channel with one sensor, achieving ((K-1)/K,1). The constructions are self-contained and the core trick—having the sensor subtract received signals across slots to cancel communication interference—is elegant. Simulation results are a useful sanity check, though the GitHub link in [41] has no URL, so the code isn't locatable.\n\nThe main soft spot is real and load-bearing. The model in Section II-A.1 says the sensing channel is 'approximately constant' over the block, but every proof uses exact equality of the sensing channel across slots. If the channel drifts, the subtraction y_s(1)-y_s(t) leaves a residual proportional to the communication signal, with power scaling as P. The claimed sDoF then collapses. The paper neither states an exact-constancy assumption for the theoretical results nor gives a drift-rate bound that would make the residual vanish. I think the intended fix is to move the model to exact block fading for the theory; the tradeoff points would survive that. This needs to be explicit.\n\nThere are also formal typos in the Theorem 4 proof: the t0 formula has the block length and sensing dimension swapped, the identity matrix size is wrong, and the text says the sensor gets U observations per block when it should be U+1. The MU-MIMO appendix has similar dimension slips. Fixable, but currently they make the proofs harder to trust than they should be.\n\nThe central argument holds up under exact channel constancy, and the flaws are in presentation and modeling precision rather than in the conceptual construction. This is for researchers in ISAC information theory and in bistatic sensing design. It deserves a serious referee: the ideas are timely, the tradeoff points are new, and a moderate revision would clean it up. I would send it to peer review with the request to reconcile the channel-constancy assumption with the proofs and fix the typos. I wouldn't cite the current version in my own work until that's done.","headline":"Genuinely new application of BIA/TIM to bistatic ISAC with clean tradeoff points, but the 'approximately constant' sensing-channel assumption doesn't match the exact constancy the proofs require, and Theorem 4's proof has dimension typos.","tokens_in":33843,"tokens_out":9653,"would_cite":false,"duration_ms":75401,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["94A15","94A40"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that adding a passive sensor to a K-user interference channel need not reduce the communication degrees of freedom, by cancelling unknown communication messages at the sensor through blind interference alignment and…","keywords":["integrated sensing and communication","bistatic ISAC","blind interference alignment","topological interference management","degrees of freedom","heterogeneous coherence times","network topology","channel estimation"],"falsifier":"Run the paper's 6-slot construction for a 3-user interference channel with a sensor, but make the sensor channel change between slot 1 and slot 2 by more than the noise floor; the difference $y_s(1)-y_s(2)$ then contains the residual term $(H_{s1}(1)-H_{s1}(2)) W_1$ alongside the intended sensing observation, so the communication symbol does not cancel. A single such measurement, or a simulation where the sensor-channel Doppler is swept while all other settings are fixed, would settle whether the claimed sDoF $2/3$ point is physically reachable.","tokens_in":32840,"feed_emoji":"📡","tokens_out":10450,"duration_ms":93566,"temperature":0.7,"pith_summary":"The paper asks whether a bistatic integrated sensing and communication (ISAC) system with separate transmitters, dispersed communication receivers, and one passive sensor can sense the environment without sacrificing communication throughput. Its answer is yes when the two links are heterogeneous: if the communication channel changes every time slot while the sensor channel stays constant over a block, blind interference alignment lets the sensor subtract the unknown communication messages from its observations; if the sensor and receivers instead have different, known connectivity patterns, topological interference management aligns those messages into a low-dimensional subspace that can be projected out. The paper derives closed-form tradeoff points between sensing degrees of freedom (sDoF, effective independent observations per slot) and communication degrees of freedom (cDoF, message symbols recovered per slot) for the K-user interference channel, MU-MISO and MU-MIMO channels, regular networks, and neighboring-antidotes networks, each strictly improving on time-sharing between sensing-only and communication-only operation. If the central theorem is right, adding a passive sensor to a K-user interference channel costs zero communication degrees of freedom while delivering $(K-1)/K$ sensing degrees of freedom.","feed_headline":"Full communication DoF survives adding a sensor to K-user channel","feed_subtitle":"Blind interference alignment and topological interference management give sensing for free by exploiting channel heterogeneity.","key_machinery":"The mechanism is a joint use of repeated communication symbols, dedicated sensing pilots known to the receiver and sensor, and channel heterogeneity. For the BIA-based schemes, the sensor forms differences $y_s(1)-y_s(t)$; because the same message symbol and the same sensor channel appear in both slots, all communication interference cancels and only the known sensing pilot $\\mathbf{h} \\mathbf{x}^0_{t-1}$ plus noise remains, producing one effective observation. For the TIM-based schemes, cyclic coding with random vectors $\\mathbf{v}^{(1)},\\ldots,\\mathbf{v}^{(K)}$ aligns all undesired messages in each receiver's and the sensor's observation into a subspace of fixed dimension, and left null vectors of that interference subspace project the interference out. In both constructions the transmitters never need the sensor channel: they exploit either the coherence-time asymmetry or the known network topology.","core_discovery":"For the bistatic ISAC model with dispersed receivers and a single-antenna sensor whose channel is unknown to the transmitters, the paper establishes that communication interference at the sensor can be cancelled perfectly rather than merely suppressed. In the K-user interference channel with heterogeneous coherence times, the point $(\\text{sDoF},\\text{cDoF}) = ((K-1)/K, 1)$ is achievable: each transmitter repeats its message over K slots while superimposing orthogonal sensing pilots, and the sensor subtracts later slots from the first, so the constant communication contribution cancels and K-1 effective channel observations remain. This point strictly improves on the time-sharing line between $(1,0)$ and $(0,1)$, which at sDoF $(K-1)/K$ would allow only cDoF $1/2$. Analogous closed-form points are proved for MU-MISO, MU-MIMO, $(K+1,U,D)$ neighboring-antidotes networks, and $(K+1,d)$ regular networks, and simulations show the schemes reduce sensing channel-estimation error by 7-48 dB over treating interference as noise and by 7-8 dB over successive interference cancellation at high SNR.","pith_inferences":["Beyond the paper: if the sensor channel is only approximately constant, drifting within the repetition block, the difference operation leaves residual communication interference proportional to the channel change, so the sDoF should degrade smoothly rather than fail abruptly; quantifying that degradation is a natural next step the paper does not take.","Beyond the paper: the BIA difference construction is not ISAC-specific, so any passive receiver with a slow channel that must coexist with a fast-varying data stream could use the same trick to convert the unknown stream into an effective training observation.","Beyond the paper: in the topological schemes, the sensor's connectivity set is tied to one of the network's receiver positions, so a sensor with an arbitrary connectivity pattern would require a new alignment construction; the paper's Remark 1 covers only the restricted cases where the sensor sees a subset of one receiver's transmitters.","Beyond the paper: the DoF analysis is asymptotic and does not optimize the power split between sensing and communication; a testable extension is to allocate power between the two under the same alignment structure to maximize finite-SNR channel-estimation accuracy."],"forward_implications":["In the K-user interference channel with one added sensor, $(\\text{sDoF},\\text{cDoF}) = ((K-1)/K, 1)$ is achievable, while time-sharing at the same sDoF gives only cDoF $1/2$.","For MU-MISO and MU-MIMO channels with $m$ transmit antennas and $K$ receivers, the achieved points $((\\lceil m/K\\rceil-1)/\\lceil m/K\\rceil, m/\\lceil m/K\\rceil)$ and $((\\lceil m/N\\rceil-1)/\\lceil m/N\\rceil, m/\\lceil m/N\\rceil)$ with $N=\\sum_k n_k$ beat time-sharing whenever the relevant ceiling exceeds one.","With only topology knowledge, the $(K+1,U,D)$ neighboring-antidotes network achieves $((U+1)/(K-D+U+1), K(U+1)/(K-D+U+1))$ and the $(K+1,d)$ regular network achieves $(2/(d+1), 2K/(d+1))$, both strictly above the corresponding time-sharing lines.","The schemes cancel communication interference at the sensor without the transmitters knowing the sensor channel, which the paper argues is the realistic regime for passive sensing nodes.","In end-to-end simulation with DQPSK and LDPC coding, the proposed schemes improve sensing channel-estimation error by roughly 7-48 dB over treating interference as noise and 7-8 dB over successive interference cancellation at high SNR, at the same asymptotic complexity order and comparable communication performance."],"supporting_citations":[{"why":"Supplies the blind interference alignment technique whose repeat-and-subtract structure cancels unknown communication symbols at the sensor.","marker":"[26]"},{"why":"Supplies the topological interference management framework and index-coding formulation used for the connectivity-based schemes.","marker":"[27]"},{"why":"Defines the neighboring-antidotes network and its communication-only DoF, the baseline that Theorem 4 improves on.","marker":"[28]"},{"why":"Defines the regular network with cooperative transmitters and gives the communication-only DoF used in Theorem 5.","marker":"[29]"},{"why":"Establishes the sDoF/cDoF tradeoff framework and the sensing-optimal and communication-optimal endpoints used for comparison.","marker":"[5]"},{"why":"Provides the bistatic ISAC decoding-and-estimation baselines, treating interference as noise and SIC, that the simulations benchmark against.","marker":"[17]"},{"why":"Supplies the neighboring-antidotes connectivity model and its achievable degrees of freedom used in the heterogeneous-connectivity construction.","marker":"[35]"}],"fun_headline_variants":["Blind IA and TIM give sensing for free in bistatic ISAC","Bistatic ISAC: Interference cancellation keeps full DoF while sensing","Sensing without sacrificing communication DoF: Blind IA and TIM in bistatic ISAC","Full cDoF preserved with (K-1)/K sDoF via blind interference alignment","Bistatic ISAC: Blind IA and TIM beat time-sharing for sensing and communication"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is a strict asymmetry: the sensing channel must stay constant over the entire transmission block while communication channels vary every slot (or, for the topology models, the connectivity pattern must be exactly known and stable); if the sensing channel drifts inside the block, the subtraction that cancels the unknown communication messages no longer works.","fun_headline_variants_meta":{"raw":{"variants":["Blind IA and TIM give sensing for free in bistatic ISAC","Bistatic ISAC: Interference cancellation keeps full DoF while sensing","Sensing without sacrificing communication DoF: Blind IA and TIM in bistatic ISAC","Full cDoF preserved with (K-1)/K sDoF via blind interference alignment","Bistatic ISAC: Blind IA and TIM beat time-sharing for sensing and communication"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00027,"raw_usage":{"total_tokens":1661,"prompt_tokens":1019,"completion_tokens":642,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":635,"completion_tokens_details":{"reasoning_tokens":534}},"tokens_in":635,"tokens_out":642,"duration_ms":5880,"temperature":1.0,"reasoning_tokens":534,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T21:54:59.040502+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the paper's 6-slot construction for a 3-user interference channel with a sensor, but make the sensor channel change between slot 1 and slot 2 by more than the noise floor; the difference $y_s(1)-y_s(2)$ then contains the residual term $(H_{s1}(1)-H_{s1}(2)) W_1$ alongside the intended sensing observation, so the communication symbol does not cancel. A single such measurement, or a simulation where the sensor-channel Doppler is swept while all other settings are fixed, would settle whether the claimed sDoF $2/3$ point is physically reachable.","supporting_citations":[{"cited_title":"Blind interference alignment,","cited_arxiv_id":null,"evidence_quote":"Supplies the blind interference alignment technique whose repeat-and-subtract structure cancels unknown communication symbols at the sensor."},{"cited_title":"Topological interference management through index coding,","cited_arxiv_id":null,"evidence_quote":"Supplies the topological interference management framework and index-coding formulation used for the connectivity-based schemes."},{"cited_title":"Index coding—an interfer- ence alignment perspective,","cited_arxiv_id":null,"evidence_quote":"Defines the neighboring-antidotes network and its communication-only DoF, the baseline that Theorem 4 improves on."},{"cited_title":"Topological interference management with transmitter cooperation,","cited_arxiv_id":null,"evidence_quote":"Defines the regular network with cooperative transmitters and gives the communication-only DoF used in Theorem 5."},{"cited_title":"Information- theoretic limits of bistatic integrated sensing and communication,","cited_arxiv_id":null,"evidence_quote":"Provides the bistatic ISAC decoding-and-estimation baselines, treating interference as noise and SIC, that the simulations benchmark against."},{"cited_title":"Elements of Cellular Blind Interference Alignment --- Aligned Frequency Reuse, Wireless Index Coding and Interference Diversity","cited_arxiv_id":"1203.2384","evidence_quote":"Supplies the neighboring-antidotes connectivity model and its achievable degrees of freedom used in the heterogeneous-connectivity construction."}],"review_version":1}