Zhenguo Gao

WDFAD-DBR

The design of routing protocols for Underwater Acoustic Sensor Networks (UASNs) poses many challenges due to long propagation, high mobility, limited bandwidth, multi-path and Doppler effect. Because of the void-hole caused by the uneven distribution of nodes and sparse deployment, the selection of next hop forwarding nodes only based on the state of current node may result in the failure of forwarding in the local sparse region. In order to reduce the probability of encountering void holes in the sparse networks, in this paper we present weighting depth and forwarding area division DBR routing protocol, called WDFAD-DBR. The novelties of WDFAD-DBR lie in: firstly, next forwarding nodes are selected according to the weighting sum of depth difference of two hops, which considers not only the current depth but also the depth of expected next hop. In this way, the probability of meeting void holes is effectively reduced. Secondly, the mechanisms for forwarding area division and neighbor node prediction are designed to reduce the energy consumption caused by duplicated packets and neighbors requests, respectively. Thirdly, we make theoretical analyses on routing performance in case of considering channel contending with respect to delivery ratio, energy consumption and average end-to-end delay. Finally we conduct extensive simulations using NS-3 simulator to verify the effectiveness and validity of WDFAD-DBR.

A novel key generation method for wireless sensor networks based on system of equations

Many key management schemes were proposed for protecting wireless sensor networks (WSNs). While applying key management to the network, it is important to ensure that the efficiency of the network is not greatly affected by key connectivity. Poor connectivity might lead to many messages forwarding. Consequently, a large amount of energy of the involved nodes would be consumed during message forwarding, which is not suitable for the resources-constraint sensor nodes. In this work, we analyze the impact of key connectivity on the efficiency of communication. Then, a novel key generation method based on system of equations is proposed to improve key connectivity of key management. The involved equations are applied to establish secret keys and each node uses these keys for protecting their communication. The system of equations is constructed to have one and only one solution so that the unique solution can also be used to establish a shared hidden key for enhancing the association among nodes. As a result, neighbors can directly communicate with each other through the shared hidden key even though they do not have common keys. To differentiate from normal keys, keys generated by the proposed method are called associated-keys. According to the analyses, we recommend that systems of linear equations (linear systems) instead of systems of polynomial equations are used to realize the proposed method with respect to the computation complexity. Furthermore, we illustrate that linear systems of two variables are sufficient to generate keys for large scale of networks. The Exclusion Basis System (EBS) is used as a instance to illustrate the implementation of key management with associated-keys. The theoretical analyses and simulation results show that key management schemes with associated-keys have better key connectivity than the corresponding schemes with normal keys. Meanwhile, other performance metrics are unaffected.

Outage Probability Equivalency of Three Typical Relay Selection Schemes for Selective DF Relay Networks with Selection Combining

Outage probability of selective decode-and-forward (DF) relay networks with different relay selection schemes are usually analyzed separately in the literature. In this paper, for selective DF relay networks with selection combining technique, we show that, the three typical relay selection schemes (named as proactive, reactive, and backward reactive relay selection scheme, respectively) are all equivalent with each other in term of outage probability of the networks. Furthermore, this outage probability equivalency is valid even when the wireless channels in the networks are non-identical, interrelated, and with different fading models. Hence, outage probability of a selective DF relay network with one particular relay selection scheme could be obtained indirectly, if necessary, by obtaining the outage probability of the network with another equivalent relay selection scheme but with lower computing complexity.

Outage Analysis of Cognitive AF Relay Networks with Multiple Primary Users and Heterogeneous Non-identical Parameters

In this paper, we investigate outage performance of dual-hop cognitive AF relay networks with multiple primary users (PUs) and heterogeneous non-identical parameters, including non-identical Rayleigh fading channels, non-identical interference power constraints of PUs, and non-identical noise powers in signals. To overcome the analysis complexity resulted from the consideration of heterogeneous non-identical parameters completely, we conduct our analysis in two steps. In the first step, we focus on networks with non-identical Rayleigh fading channels but with identical interference power constraints and identical noise powers. As analysis results, both the exact closed-form expression and asymptotic closed-form expression of the cumulative distribution function of equivalent end-to-end signal-to-noise ratio of the network are obtained. Then in the second step, we propose a method for transforming networks with heterogeneous non-identical parameters into new networks with identical interference power constraints and identical noise powers meanwhile guaranteeing that outage performance of the two networks before and after the transformation are identical. Hence, outage performance of networks with heterogeneous non-identical parameters can be obtained by applying the analysis results in the fist step to the corresponding transformed networks. Numerical simulations validate the correctness of our analysis, and the simulations also indicate that taking multiple PUs and the diversity of the parameters into consideration is of key importance in obtaining a more accurate estimation of outage performance of such networks.

KM-based efficient algorithms for optimal packet scheduling problem in celluar/infostation integrated networks

Abstract Request models in previous works on the Packet Scheduling (PS) problem in cellular/infostation integrated networks usually have some restrictions which make them less practical in some situations. In this paper, we investigate the packet scheduling problem with our new request model, which is general enough to include previous request models as special cases. We first propose a Greedy algorithm for the PS problem and analyze its approximation ratio. We then transform our PS problem to the mini-slot mini-request version PS problem (MSMR-PS problem) by using the concept of mini-slot and mini-request. Next, by embedding problem information into bipartite graphs, we investigate the relationships between the MSMR-PS problem and the corresponding maximum matching (MM) problem on these graphs. Based on these relations, we propose efficient algorithms to obtain quasi-optimal solutions to the original PS problem by using the Kuhn–Munkres (KM) algorithm to solve the MM problems on the bipartite graphs. These algorithms include two offline algorithms (named as KMUpper and KMLower) and one online algorithm named KMOnline. KMUpper and KMLower return upper and lower bounds for optimal profit of the PS problem, respectively. Simulation results show that our algorithms are more preferable than the other existing algorithms like FIFO, exponential capacity algorithm.

Efficient key generation leveraging wireless channel reciprocity for MANETs

Abstract Wireless channel reciprocity ensures that transceivers can generate the shared secret key from their communicating channels. However, various interferences in real environments make channel measurement sequences of different transceivers are highly correlated accompanying with many discrepancies. Then, if these measurements are directly quantified (and encoded) to bit sequences, the bit mismatch rate might be extremely high. Consequently, the efficiency and security of key generation might be reduced since some information might be leaked during key agreement. In this paper, an efficient key generation scheme is proposed, which is available to generate the shared secret key for transceivers even facing the referred challenge. During key generation, curve fitting is implemented to pre-process measurements so that the reciprocity of outputs of curve fitting processes are better than original channel measurements. Then, the bit mismatch rate can be greatly reduced when these outputs are used to generate bit sequences. To extract the shared bit sequence, the interactive Cascade is implemented during information reconciliation. Finally, 2-universal hash functions are used to achieve privacy amplification so that the leaked information can be eliminated and randomness of the shared key can be guaranteed. Several experiments in real mobile scenarios are conducted to evaluate the proposed scheme, where the received signal strength measurements are derived from heterogeneous devices. The results show that the implementation of pre-processing can effectively reduce the bit mismatch rate and the proposed scheme can efficiently generate shared secret keys for transceivers in real environments.

An Optimal Algorithm for the One-Dimensional RSU Deployment Problem with a New Model

Proper deployment of roadside units (RSUs) is of crucial importance to vehicular ad hoc networks (VANETs). However, our understandings to the simple one-dimensional RSU Deployment (D1RD) problem with nonuniform profit density is still seriously limited. In this paper, we analyze the D1RD problem and try to design optimal algorithms for it. We first analyze the properties of the optimal solutions of the D1RD problem involving a single RSU, and then extend to multiple RSUs. Next, we propose an efficient technique named Dynamic Limiting (DynLim), which reduces the solution search space size considerably by adjusting search space limits dynamically. Finally, an optimal algorithm named OptDynLim is proposed based on the DynLim technique, and its optimality is proved. Numerical simulations validate the correctness of our analyzes and show that DynLim can usually reduce solution search space size by more than 99%.

Outage Probability of Cognitive Selective DF Relay Networks with Multiple Primary Nodes and Heterogenous Non-Identical Constraints

Outage probability analysis of cognitive selective DF relay networks with selection combining technique taking into account all the following issues is important: multiple SU relays, multiple PUs, non-identical independent Rayleigh fading channels, non-identical interference power limits of PUs, non-identical maximum transmission power limits of SUs, and non-identical noise powers in signals. However, no existing works solve this task exactly. In this paper, to overcome the complexity in analyzing such non-identical-parameter networks directly, we conduct the analysis in two steps indirectly. In the first step, we obtain outage probability expressions of identical-parameter networks. Then in the second step, we propose a method for transforming a non-identical-parameter network into a new identical-parameter network, meanwhile guaranteeing that the distributions of end-to-end signal to noise ratio of the two networks before and after the transformation are identical. Thus the outage probability of the identical-parameter network can be regarded as outage probability of the corresponding non-identical-parameter network. Numerical simulations validate our analysis results and confirm the feasibility of the two-step indirect analysis approach.

Constructing Longer Sequences Using a New Improved Interleaved Structure

In this letter, a new interleaved structure is improved and its interesting Local Coordinated Shifting (LCS) Property which is helpful in constructing longer sequences is emphasized. Then, making use of the interleaved structure, several methods for constructing perfect ternary sequences and three-level autocorrelation (0, 1) binary sequences are proposed. The LCS property and the demonstrational construction methods imply that this new interleaved structure is an interesting approach for constructing new sequences.