An Information Theoretic Analysis of Ghost Modulation

Daniel Harman, Ashton Palacios, Philip Lundrigan, Willie K. Harrison · 2024

Read An Information Theoretic Analysis of Ghost Modulation by Daniel Harman, Ashton Palacios, Philip Lundrigan, Willie K. Harrison (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-05249.txt.

Introduction

Daniel Harman and co-authors lay the math base for a scheme that hides a slow message inside a normal network stream. Their 2024 paper calls the method Ghost Modulation. It rides on cross-technology links, where one kind of gear talks through another's signals, and it needs only a software update to run.

Such a channel is lopsided: packets drop and delays blur the timing, which makes the math tricky. The authors write out models for the signal and the channel, then give a rule for reading the message that follows best-guess logic. They add a method to find the packets and lock onto their timing, backed by simulations. Known short error-correcting codes are then applied, with error rates reported.

Catalog record

CREATORDaniel Harman, Ashton Palacios, Philip Lundrigan, Willie K. Harrison
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE27 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-05249

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

An Information Theoretic Analysis of Ghost Modulation Thanks: This work was partially funded by the US National Science Foundation: Grant Award Numbers #2135732, #2153317, and #2030165. Daniel Harman, Ashton Palacios, Philip Lundrigan, Willie K. Harrison Abstract Side channels have become an essential component of many modern information-theoretic schemes. The emerging field of cross technology communications (CTC) provides practical methods for creating intentional side channels between existing communications technologies. This paper describes a theoretical foundation for one such, recently proposed, CTC scheme: Ghost Modulation (GM). Designed to modulate a low-data-rate message atop an existing network stream, GM is particularly suited for transmitting identification or covert information. The implementation only requires firmware updates to existing hardware, making it a cost-effective solution. However, GM provides an interesting technical challenge due to a highly asymmetric binary crossover erasure channel (BCEC) that results from packet drops and network delays. In this work, we provide a mathematical description of the signal and channel models for GM. A heuristic decision rule based on maximum-likelihood principles for simplified channel models is proposed. We describe an algorithm for GM packet acquisition and timing synchronization, supported by simulation results. Several well known short block codes are applied, and bit error rate (BER) results are presented. I Introduction Side channels have become an important topic within information theory in recent years because of their implications on security [ 1 ] and covertness [ 2 , 3 ] . They will likely continue to be an important topic moving forward into the era of post-quantum cryptography, with concerns over digital privacy. Most of the existing literature has centered on mitigating information leakage from unintended side channels to preserve security and privacy. This is not always the case, however, as intended side channels can be used for covert communications. Cross technology communications (CTC) is an emerging field in communications and networking theory that enables devices not explicitly designed to communicate with each other to share information, often at ultra-low bit rates. The schemes create intentional side channels, allowing existing communication infrastructures to adapt to new technologies and address unforeseen challenges. Recent CTC research has focused on smart devices and the internet of things (IoT) to facilitate simple, yet crucial tasks such as 802.11 Wi-Fi to 802.15.4 Zigbee communication [ 4 ] , improving smart device setup processes [ 5 ] or enabling long-range IoT sensor updates from 802.11 devices by modulating the noise floor [ 6 ] . Fig. 1: Ghost modulation symbol structure. Ghost Modulation (GM) [ 7 ] , a recently proposed CTC protocol, is designed to transmit secondary messages over an existing network for applications like embedding identifying information in satellite internet networks. GM introduces artificial delays into the network stream, treating each packet as a pulse, to communicate the secondary message. This technique is conceptually similar to binary pulse position modulation (PPM). GM is capable of being implemented at any level of the open systems interconnection (OSI) model [ 8 ] , with a particular emphasis on the higher layers, such as the application layer. GM presents several intriguing technical challenges. A key challenge is that unlike traditional modulations with strict hardware-based timing control, GM operates without such precision. GM is vulnerable not only to normal channel noise, but the network hops themselves will introduce delays into the packet timing. These delays introduce a second major error mode, distinct from those encountered in typical channels. Even in single-hop networks, packets can be delayed within the operating system or by the network interface controller (NIC). In this work, we present a model for this unusual asymmetric channel and analyze its capacity. Additionally, while robust receiver time synchronization schemes exist for major PPM variations [ 9 , 10 , 11 , 12 , 13 , 14 , 15 ] , the network-induced delays that GM encounters differ significantly from traditional timing issues such as clock offsets and jitter. Consequently, existing literature on PPM time synchronization is insufficient to address GM’s challenges, necessitating a novel joint acquisition and timing synchronization scheme. In this work, we apply information-theoretic analysis to the GM signal and channel models and present detection and synchronization algorithms. Section II presents and justifies the signal model, the discrete channel model, and channel capacity. Simplified maximum-likelihood decision rules are derived in Section III , forming the basis of a heuristic decision rule. Section IV presents a signal acquisition and timing estimation algorithm. Finally, in Section V we apply several well-known, short-block-length, forward error correction (FEC) codes and present the resulting bit error rate (BER) performance. II Signal and Channel Models Fig. 2: Binary crossover erasure channel with fully-asymmetric transition probabilities GM operates similarly to PPM, where the pulses are packets in the network. PPM works by transmitting a pulse within one of M time-domain transmission slots of length T_{\text{slot}} that are uniformly spaced within a symbol time, T_{\text{sym}} . GM also has guard periods of length T_{\text{g}} after symbol periods, as shown in Fig. 1 . In the implementation of GM, many practical factors must be considered, such as how to identify network packets acting as pulses, how to maintain covertness, or how to best preserve network throughput. While important, this work instead focuses on the theoretical foundation of GM and, as such, we assume an equivalent signal model for binary GM. The equivalent continuous-time transmitted signal is modeled as s(t)=\sum_{k=0}^{N-1}\delta(t-kT_{\text{sym}}-\theta_{k}T_{\text{slot}}) (1) where N is the total number of symbols, T_{\text{sym}} is the symbol time, T_{\text{slot}} is the slot time for a “0” or a “1” time slot, and \theta_{k}\in\{0,1\} is the k th GM symbol. A guard window occurs after the two pulse windows, of length T_{\text{g}} , such that T_{\text{sym}}=2T_{\text{slot}}+T_{\text{g}}. (2) The choice of a Dirac delta function to represent each pulse is natural in that (1) packets are not allowed to be directly modified, and (2) all the GM information is contained in the start time of the packet. Additionally, the delta functions allow for easier modeling as they can easily be convolved with envelope functions to better represent the network packets. As GM can be built atop any packetized network, this work does not assume a specific packet type or envelope for generality’s sake. A transmitted signal may also be represented in vector form as \bm{s}=\begin{bmatrix}\theta_{0}T_{\text{slot}}\\ T_{\text{sym}}+\theta_{1}T_{\text{slot}}\\ \vdots\\ (N-1)T_{\text{sym}}+\theta_{N-1}T_{\text{slot}}\end{bmatrix}^{\intercal} (3) where each element is the time stamp of intended transmission. The continuous-time received signal is represented by r(t)=\sum_{k=0}^{N-1}z_{k}\delta(t-kT_{\text{sym}}-\theta_{k}T_{\text{slot}}-\tau_{k}) (4) where \tau_{k}\sim\text{Exp}(\lambda) is the network-induced delay, \frac{1}{\lambda} is the mean delay time, z_{k}\sim\text{Bernoulli}(1-\rho) , and \rho is the packet drop rate of the network. The received vector is \bm{r}=(\bm{s}+\bm{\tau})P^{(N)} (5) where P^{(N)} is an N\times\tilde{N} quasi-permutation matrix with 0\leq\tilde{N}\leq N. (6) P^{(N)} sorts the values of (\bm{s}+\bm{\tau}) to account for their time-ordered reception and for the possibility of dropping packets, and as such, \bm{r} is a 1\times\tilde{N} vector while \bm{s} is a 1\times N vector. II-A Discrete Memoryless Channel Model The discrete memoryless channel model in general is a binary crossover erasure channel (BCEC) 1 1 1 The more common analog of this channel is the binary symmetric erasure channel (BSEC) which features symmetric probabilities of bit-flips and symmetric erasure probabilities. depicted in Fig. 2 . We denote the transition probabilities from input state n to output

Read the complete text