In industry, automatic speech recognition has come to be a competitive feature for embedded products with poor hardware resources. In this work, we propose a tiny end-to-end speech recognition model that is lightweight and easily deployable on edge platforms. First, instead of sophisticated network structures, such as recurrent neural networks, transformers, etc., the model we propose mainly uses convolutional neural networks as its backbone. This ensures that our model is supported by most software development kits for embedded devices. Second, we adopt the basic unit of MobileNet-v3, which performs well in computer vision tasks, and integrate the features of the hidden layer at different scales, thus compressing the number of parameters of the model to less than 1 M and achieving an accuracy greater than that of some traditional models. Third, in order to further reduce the CPU computation, we directly extract acoustic representations from 1-dimensional speech waveforms and use a self-supervised learning approach to encourage the convergence of the model. Finally, to solve some problems where hardware resources are relatively weak, we use a prefix beam search decoder to dynamically extend the search path with an optimized pruning strategy and an additional initialism language model to capture the probability of between-words in advance and thus avoid premature pruning of correct words. In our experiments, according to a number of evaluation categories, our end-to-end model outperformed several tiny speech recognition models used for embedded devices in related work.

As the active safety of vehicles has become essential, vehicular communication has been gaining attention. The IETF IPWAVE working group has proposed the shared prefix model-based vehicular link model. In the shared prefix model, a prefix is shared among RSUs to prevent changes in IPv6 addresses of a vehicle within a shared prefix domain. However, vehicle movement must be tracked to deliver packets to the serving RSU of the vehicle within a shared prefix domain. The Identifier/Locator Separation Protocol (ILSP) is one of the techniques used to handle vehicle movement. It has several drawbacks such as the inability to communicate with a standard IPv6 module without special components and the requirement to pass signaling messages between end hosts. Such drawbacks severely limit the service availability for a vehicle in the Internet. We propose an ILSP for a shared prefix model over IEEE WAVE IPv6 networks. The proposed protocol supports IPv6 communication between a standard IPv6 node in the Internet and a vehicle supporting the proposed protocol. In addition, the protocol hides vehicle movement within a shared prefix domain to peer hosts, eliminating the signaling between end hosts. The proposed protocol introduces a special NDP module based on IETF IPWAVE vehicular NDP to support vehicular mobility management within a shared prefix domain and minimize link-level multicast in WAVE networks.

Memory-efficient Internet Protocol (IP) lookup with high speed is essential to achieve link-speed packet forwarding in IP routers. The rapid growth of Internet traffic and the development of optical link technologies have made IP lookup a major performance bottleneck in core routers. In this paper, we propose a new IP route lookup architecture based on hardware called Prefix-Route Trie (PR-Trie), which supports both IPv4 and IPv6 addresses. In PR-Trie, we develop a novel structure called Overlapping Hybrid Trie (OHT) to perform fast longest-prefix-matching (LPM) based on Multibit-Trie (MT), and a hash-based level matching query used to achieve only one off-chip memory access per lookup. In addition, the proposed PR-Trie also supports fast incremental updates. Since the memory complexity in MT-based IP lookup schemes depends on the level-partitioning solution and the data structure used, we develop an optimization algorithm called Bitmap-based Prefix Partitioning Optimization (BP2O). The proposed BP2O is based on a heuristic search using Ant Colony Optimization (ACO) algorithms to optimize memory efficiency. Experimental results using real-life routing tables prove that our proposal has superior memory efficiency. Theoretical performance analyses show that PR-Trie outperforms the classical Trie-based IP lookup algorithms.

The volume of digital information is growing at an extremely fast pace which, in turn, exacerbates the need of efficient mechanisms to find the presence of a pattern in an input text or a set of input strings. Combining the processing power of Graphics Processing Unit (GPU) with matching algorithms seems a natural alternative to speedup the string-matching process. This work proposes a Parallel Rabin-Karp implementation (PRK) that encompasses a fast-parallel prefix-sums algorithm to maximize parallelization and accelerate the matching verification. Given an input text T of length n and p patterns of length m, the proposed implementation finds all occurrences of p in T in O(m+q+n/τ+nm/q) time, where q is a sufficiently large prime number and τ is the available number of threads. Sequential and parallel versions of the PRK have been implemented. Experiments have been executed on p≥1 patterns of length m comprising of m=10, 20, 30 characters which are compared against a text string of length n=227. The results show that the parallel implementation of the PRK algorithm on NVIDIA V100 GPU provides speedup surpassing 372 times when compared to the sequential implementation and speedup of 12.59 times against an OpenMP implementation running on a multi-core server with 128 threads. Compared to another prominent GPU implementation, the PRK implementation attained speedup surpassing 37 times.

This paper presents study results regarding the analytical method for performance characteristics and application scheme, which cover a feature detection scheme using a Cyclic Prefix (CP) that is attached to an OFDM signal. The detection scheme is especially important when used as a sensing technology in advanced systems such as Device-to-Device (D-to-D) or Internet of Things (IoT). Herein, we present several basic performance characteristics of the signal processing involved in feature detection, namely, the Output S/N (Signal-to-Noise power ratio) and probability density functions of the OFDM signal and the noise measured at the output of the feature detector. The Output S/Nis described by an analytical expression and is also examined by conducting a software simulation. An analytical approach is investigated by modeling the spectral density of the OFDM signal and input noise and by executing the mathematical operations such as convolutional integration on the combination of OFDM signal and noise. The analytical results coincide closely with the simulation results. As for the applications to mobile communication system, some methods of the feature detection schemes are addressed. These are an estimation method for the Input C/N (Carrier-to-Noise power ratio) and a system discrimination scheme, especially under the assumption that two OFDM systems using different CP lengths are simultaneously operated in the same frequency. Furthermore, under the condition that two OFDM signals are transmitted in an asynchronous manner, a scheme to estimate their timing offset and signal power ratio is also described.

This paper shows a method to represent interval functions by using head-tail expressions. The head-tail expressions represent greater-than GT(X:A) functions, less-than LT(X:B) functions, and interval functions IN0(X:A,B) more efficiently than sum-of-products expressions. Let n be the number of bits to represent the largest value in the interval (A,B). This paper proves that a head-tail expression (HT) represents an interval function with at most n words in a ternary content addressable memory (TCAM) realization. It also shows the average numbers of factors to represent interval functions by HTs for up to n=16, which were obtained by a computer simulation. It also conjectures that, for sufficiently large n, the average number of factors to represent n-variable interval functions by HTs is at most 2/3n-5/9. Experimental results also show that, for n≥10, to represent interval functions, HTs require at least 20% fewer factors than MSOPs, on the average.

This paper presents two types of cryptanalysis on a Merkle-Damgård hash based MAC, which computes a MAC value of a message M by Hash(K||l||M) with a shared key K and the message length l. This construction is often called LPMAC. Firstly, we present a distinguishing-H attack against LPMAC instantiated with any narrow-pipe Merkle-Damgård hash function with O(2n/2) queries, which indicates the incorrectness of the widely believed assumption that LPMAC instantiated with a secure hash function should resist the distinguishing-H attack up to 2n queries. In fact, all of the previous distinguishing-H attacks considered dedicated attacks depending on the underlying hash algorithm, and most of the cases, reduced rounds were attacked with a complexity between 2n/2 and 2n. Because it works in generic, our attack updates these results, namely full rounds are attacked with O(2n/2) complexity. Secondly, we show that an even stronger attack, which is a powerful form of an almost universal forgery attack, can be performed on LPMAC. In this setting, attackers can modify the first several message-blocks of a given message and aim to recover an internal state and forge the MAC value. For any narrow-pipe Merkle-Damgård hash function, our attack can be performed with O(2n/2) queries. These results show that the length prepending scheme is not enough to achieve a secure MAC.

The main contribution of this paper is to show optimal parallel algorithms to compute the sum, the prefix-sums, and the summed area table on two memory machine models, the Discrete Memory Machine (DMM) and the Unified Memory Machine (UMM). The DMM and the UMM are theoretical parallel computing models that capture the essence of the shared memory and the global memory of GPUs. These models have three parameters, the number p of threads, and the width w of the memory, and the memory access latency l. We first show that the sum of n numbers can be computed in $O({nover w}+{nlover p}+llog n)$ time units on the DMM and the UMM. We then go on to show that $Omega({nover w}+{nlover p}+llog n)$ time units are necessary to compute the sum. We also present a parallel algorithm that computes the prefix-sums of n numbers in $O({nover w}+{nlover p}+llog n)$ time units on the DMM and the UMM. Finally, we show that the summed area table of size $sqrt{n} imessqrt{n}$ can be computed in $O({nover w}+{nlover p}+llog n)$ time units on the DMM and the UMM. Since the computation of the prefix-sums and the summed area table is at least as hard as the sum computation, these parallel algorithms are also optimal.

Content-Centric Networking (CCN) employs a hierarchical but location independent content naming scheme. While such a location independent naming brings various benefits including efficient content delivery, mobility, and multihoming, location independent name prefixes are hard to aggregate. This poses a serious scaling issue on the efficiency of looking up content names in a huge Forwarding Information Base (FIB) by longest prefix matching, which requires seeking the longest matching prefix through all candidate prefix lengths. We propose a new scheme for efficiently looking up non-aggregatable name prefixes in a large FIB. The proposed scheme is based on the observation that the bottleneck of FIB lookup is the random accesses to the high-latency off-chip DRAM for prefix seeking and this can be reduced by exploiting the information on the longest matching prefix length in the previous hop. Our evaluation results show that the proposed scheme significantly improves FIB lookup latency with a reasonable traffic parameters observed in today's Internet.

We propose a low-overhead fault-secure parallel prefix adder. We duplicate carry bits for checking purposes. Only one half of normal carry bits are compared with the corresponding redundant carry bits, and the hardware overhead of the adder is low. For concurrent error detection, we also predict the parity of the result. The adder uses parity-based error detection and it has high compatibility with systems that have parity-based error detection. We can implement various fault-secure parallel prefix adders such as Sklansky adder, Brent-Kung adder, Han-Carlson adder, and Kogge-Stone adder. The area overhead of the proposed adder is about 15% lower than that of a previously proposed adder that compares all the carry bits.

First, this paper derives the prefix sum-of-products expression (PreSOP) and the number of products in a PreSOP for an interval function. Second, it derives Ψ(n,τp), the number of n-variable interval functions that can be represented with τp products. Finally, it shows that more than 99.9% of the n-variable interval functions can be represented with ⌈ n - 1 ⌉ products, when n is sufficiently large. These results are useful for a fast PreSOP generator and for estimating the size of ternary content addressable memories (TCAMs) for packet classification.

In OFDM systems, the correlation of cyclic prefix (CP) with its corresponding part at the end of the symbol can be used to estimate the maximum Doppler spread. However, the estimation accuracy of this CP based method is seriously affected by the inter-symbol interference (ISI) generated in the multipath channel. In this letter, we propose an enhanced CP based method which is immune to the ISI and can hence obtain an unbiased estimate of the auto-correlation function in multipath channels.

In time-varying frequency selective channels, to obtain high-rate joint time-frequency diversity, linear dispersion coded orthogonal frequency division multiplexing (LDC-OFDM), has recently been proposed. Compared with OFDM systems, single-carrier systems may retain the advantages of lower PAPR and lower sensitivity to carrier frequency offset (CFO) effects, which motivates this paper to investigate how to achieve joint frequency and time diversity for high-rate single-carrier block transmission systems. Two systems are proposed: linear dispersion coded cyclic-prefix single-carrier modulation (LDC-CP-SCM) and linear dispersion coded zero-padded single-carrier modulation (LDC-ZP-SCM) across either multiple CP-SCM or ZP-SCM blocks, respectively. LDC-SCM may use a layered two-stage LDC decoding with lower complexity. This paper analyzes the diversity properties of LDC-CP-SCM, and provides a sufficient condition for LDC-CP-SCM to maximize all available joint frequency and time diversity gain and coding gain. This paper shows that LDC-ZP-SCM may be effectively equipped with low-complexity minimum mean-squared error (MMSE) equalizers. A lower complexity scheme, linear transformation coded SCM (LTC-SCM), is also proposed with good diversity performance.

Prefix caching improves the performance of IP lookup by exploiting spatial and temporal locality of IP references. However, it cannot cache non-leaf prefixes, so several prefix expansion schemes have been proposed to handle those prefixes. Such schemes have some drawbacks to incur modification of routing table or severe miss penalty. We propose an efficient prefix expansion scheme which achieves good performance without additional burden to lookup scheme. In the proposed scheme a non-leaf prefix is expanded to the length of the longest immediate descendant prefix when it is cached. Evaluation result shows our scheme achieves very low miss ratio even though it does not increase the size of routing table and cache miss penalty.

High-speed IP address lookup with fast prefix update is essential for designing wire-speed packet forwarding routers. The developments of optical fiber and 100 Gbps interface technologies have placed IP address lookup as the major bottleneck of high performance networks. In this paper, we propose a novel structure named Compressed Multi-way Prefix Tree (CMPT) based on B+ tree to perform dynamic and scalable high-speed IP address lookup. Our contributions are to design a practical structure for high-speed IP address lookup suitable for both IPv4 and IPv6 addresses, and to develop efficient algorithms for dynamic prefix insertion and deletion. By investigating the relationships among routing prefixes, we arrange independent prefixes as the search indexes on internal nodes of CMPT, and by leveraging a nested prefix compression technique, we encode all the routing prefixes on the leaf nodes. For any IP address, the longest prefix matching can be made at leaf nodes without backtracking. For a forwarding table with u independent prefixes, CMPT requires O(logmu) search time and O(mlogmu) dynamic insert and delete time. Performance evaluations using real life IPv4 forwarding tables show promising gains in lookup and dynamic update speeds compared with the existing B-tree structures.

Multipath routing has been extended to Border Gateway Protocol (BGP), the current de facto inter-domain routing protocol, to address the reliability and performance issues of the current Internet. However, inter-domain multipath routing introduces a significant challenge for scalability due to the large scale of the inter-domain routing system. At the same time it also introduces new challenges in terms of security and security related overhead. In this paper, we propose a regional multipath approach, Regional Multipath Inter-domain Routing (RMI), where multiple paths are only allowed to be propagated within a well-defined range. With multipath routing in a region, we enable inter-domain routing with rich path diversity and improved security, and no longer have to sacrifice scalability. We show how to propagate multiple paths based on the region by theoretical analysis and by extensive simulations. Our simulations show that the number of messages generated using this approach and the convergence delay are much less than those of BGP and BGP with full multipath advertisement.

Overcoming the highly organized and coordinated malware threats by botnets on the Internet is becoming increasingly difficult. A honeypot is a powerful tool for observing and catching malware and virulent activity in Internet traffic. Because botnets use systematic attack methods, the sequences of malware downloaded by honeypots have particular forms of coordinated pattern. This paper aims to discover new frequent sequential attack patterns in malware automatically. One problem is the difficulty in identifying particular patterns from full yearlong logs because the dataset is too large for individual investigations. This paper proposes the use of a data-mining algorithm to overcome this problem. We implement the PrefixSpan algorithm to analyze malware-attack logs and then show some experimental results. Analysis of these results indicates that botnet attacks can be characterized either by the download times or by the source addresses of the bots. Finally, we use entropy analysis to reveal how frequent sequential patterns are involved in coordinated attacks.

Although IP address lookup schemes using ternary content addressable memory (TCAM) can perform high speed packet forwarding, TCAM is much more expensive than ordinary memory in implementation cost. As a low-cost solution, binary search algorithms such as a binary trie or a binary search tree have been widely studied. This paper proposes an efficient IP address lookup scheme using balanced binary search with minimal entries and optimal prefix vectors. In the previous scheme with prefix vectors, there were numerous pairs of nearly identical entries with duplicated prefix vectors. In our scheme, these overlapping entries are combined, thereby minimizing entries and eliminating the unnecessary prefix vectors. As a result, the small balanced binary search tree can be constructed and used for a software-based address lookup in small-sized routers. The performance evaluation results show that the proposed scheme offers faster lookup speeds along with reduced memory requirements.

In this letter we present efficient iterative frequency domain equalization for single-carrier (SC) transmission systems with insufficient cyclic prefix (CP). Based on minimum mean square error (MMSE) criteria, iterative decision feedback frequency domain equalization (IDF-FDE) combined with cyclic prefix reconstruction (CPR) is derived to mitigate inter-symbol interference (ISI) and inter-carrier interference (ICI). Computer simulation results reveal that the proposed scheme significantly improves the performance of SC systems with insufficient CP compared with previous schemes.

In this paper, we propose a simple but effective way of improving the performance of channel estimation (CE) for pilot cyclic prefixed single carrier (PCP-SC) system. The proposed method utilizes the property that the shifting signal of the PCP pilot signal can also be utilized to estimate the channel information. The receiver can continuously estimate the channel information by just shifting the received pilot signal. Regardless of the signal-to-noise ratio (SNR) and the pilot type, the proposed method can achieve about a 1.72 dB performance gain in terms of the mean squared error (MSE) of channel estimation with a slight increase in computational complexity. The BER performance with the proposed CE improvement are evaluated in a multipath fading channel using a zero-forcing (ZF) equalizer and an minmum mean squared error (MMSE) equalizer by computer simulation. It is shown that the proposed CE improvment method using an MMSE equalizer which has an unbiased vlaue of noise variance (NV) estimator gives a promising BER performance. The proposed method also benefits the estimation of the SNR for the single carrier system.