认知无线电通信网络：探索无线通信新边界

需积分: 9 26 浏览量更新于2024-07-31 收藏 10.29MB PDF 举报

"cognitive wireless communication networks" 认知无线通信网络是一种先进的无线通信技术，它旨在最大化有限的无线电频谱利用效率，以应对无线网络中不断增长的服务和应用需求。本书《Cognitive Wireless Communication Networks》由Ekram Hossain和Vijay Bhargava两位通信领域的专家编辑，详细探讨了这一技术的各个方面。认知无线电的核心在于其智能化的能力，它能够感知、理解和适应环境变化。这种技术允许设备动态地检测和利用未被授权但空闲的频谱，从而提高频谱效率，减少干扰，并提升整体网络性能。认知无线电网络主要包含以下几个关键技术： 1. **频谱感知**：这是认知无线电的基础，设备通过检测当前频谱使用情况来确定哪些频段可以安全地使用。这通常涉及复杂的信号处理算法，如能量检测、特征检测和盲源分离等。 2. **频谱决策**：在获取频谱使用信息后，认知无线电需要决定何时、何处以及如何接入空闲频段。这涉及到策略制定，如避免与授权用户的冲突，确保服务质量（QoS）和最小化干扰。 3. **频谱共享**：认知无线电必须有效地与授权用户共享频谱，这可能包括动态频谱接入（DSA）、频谱预留、多频段操作等多种方式。 4. **学习与适应**：认知无线电系统通过机器学习和自适应算法来不断学习和改进其操作策略，以适应不断变化的无线环境。 5. **网络协作**：在认知无线网络中，节点之间进行信息交换和协作，以优化网络性能，例如，通过分布式算法实现更有效的频谱管理。 6. **安全与隐私**：由于认知无线电的动态特性，它对安全性和隐私保护提出了新的挑战。这包括防止恶意攻击，保护用户数据，以及确保网络的可靠性和稳定性。 7. **标准与政策**：随着认知无线电技术的发展，国际标准组织和监管机构正在制定相关规定，以促进其广泛应用并解决可能的法律和政策问题。 8. **应用实例**：认知无线电技术已在多个领域找到应用，如军事通信、应急服务、物联网（IoT）以及未来的5G移动通信系统。《Cognitive Wireless Communication Networks》这本书深入讨论了这些主题，并提供了理论分析、实验结果和实际应用案例，对于研究人员、工程师和学生来说，是理解认知无线电网络原理和技术的宝贵资源。它不仅涵盖了理论基础，还探讨了实施中的挑战和解决方案，有助于推动这一领域的进一步发展。

XVIII Preface

suboptimal solution to the problem is then proposed which reduces the computational

complexity signiﬁcantly with only moderate loss in performance.

The essence of the optimal sensing and access policy is that it achieves a tradeoff

among gaining instantaneous reward, gaining information for future use, and conser-

vation of energy. Speciﬁcally, the optimal sensing decision (i.e., to sense or not to

sense) strikes a balance between gaining reward and conserving energy. The optimal

access policy depends on the energy consumption due to sensing and the channel

fading conditions. If the energy consumption due to sensing is negligible, under poor

channel conditions, a cognitive radio should wait for the best channel condition. On

the other hand, if the sending energy consumption is large, it should access the chan-

nel regardless of the fading condition.

Decentralized Dynamic Spectrum Access: Game Theoretic Learning Model

As has been stated before, game theory provides useful models for developing dy-

namic spectrum access techniques which involve multiple competing cognitive

ra-

dios. Chapter 11, authored by Michael Maskery,

Vikram Krishnamurthy, and Qing

Zhao, presents a game theoretic learning model for dynamic spectrum access in a

distributed cognitive radio environment. Speciﬁcally, a decentralized game theoretic

reinforcement learning scheme is proposed for slotted carrier sense multiple access

(CSMA)-based channel access for cognitive radios. Each radio competes for oppor-

tunistic access to a number of channels used by the primary users. The objective of

each radio is to select a subset of the unoccupied channels to satisfy its current de-

mand level. Each cognitive radio aims at maximizing its utility which is formulated

as a function of its demand level, channel usage pattern of primary users, number of

competing users for a given channel, pricing for the available radio channels, and the

throughput achieved. The utility function is used to guide a modiﬁed regret-based re-

inforcement learning procedure. This decentralized procedure allows each cognitive

radio to perform channel access satisfactorily through repeated channel selections

and corresponding performance measurements. Regret-based learning procedures

are simple to implement and their convergence properties are well understood.

The proposed learning algorithm is game theoretic in nature since each cogni-

tive radio adopts a channel allocation that maximizes its own utility in response to

the actions by other cognitive radios (i.e., competitive optimality). In a regret-based

learning algorithm, a regret matrix is used by each user, which tracks for every pair

of actions a

the difference in utility if the user had taken action a

in the past

everywhere he took action a

. Given that the action taken by the user at time instant

n is a

, the probability of choosing action a

at time instant n +1is proportional

to the regret value from a

to a

. The learning process proceeds through exploration

by switching among different actions. Since the algorithm requires that a use knows

the utility he would have received for each action even if that action was not taken,

which might not be feasible in practice, a stochastic approximation is used to com-

pute the regret values. The modiﬁed regret-based learning algorithm converges to a

set of correlated equilibria. As we described before, correlated equilibrium is a more

general equilibrium concept than Nash equilibrium.

Preface XIX

The performance of a decentralized spectrum access scheme may be improved by

occasionally adjusting the behavior of the cognitive radios, for example, by periodi-

cally broadcasting parameter (e.g., spectrum price) updates from a central controller.

These updates can be used by the cognitive radios to update their utility function.

This adjustment may improve the equilibrium behavior from a global perspective,

for example, improve the spectral efﬁciency of the cognitive radio system (i.e., aver-

age proportion of radio channels used by the cognitive radios). The pricing parameter

should be determined such that it maximizes the global utility which is a function of

the equilibrium behavior of the cognitive radios under the utility determined as a

function of this pricing parameter. Since it is difﬁcult (if not impossible) to deter-

mine the equilibrium behavior a priori, a stochastic approximation approach is used

to ﬁnd the optimal value of the pricing parameter.

Decentralized Dynamic Spectrum Access in Cognitive Networks: A Local

Coordination/Bargaining-Based Approach

Chapter 12, authored by Haitao Zheng and Lili Cao, presents a decentralized ap-

proach for spectrum sharing where the cognitive users perform local coordinations

to modify their spectrum usage (to adapt to the variations in network topology) to

achieve a conﬂict-free spectrum assignment which maximizes the system utility. In

the considered system model, cognitive radios select communication channels and

adjust transmission power accordingly so that they do not cause interference to the

primary users. The system utility is measured in terms of proportional fairness (i.e.,

total logarithmic user throughput) among the cognitive radios. For local coordina-

tion, the cognitive radios consider both the needs of neighboring devices and spec-

trum availability to determine their spectrum usage. Upon detection of suboptimal

spectrum usage (e.g. due to user mobility), the coordination/bargaining protocol is

triggered to apply local adjustments among the neighboring radios. During local co-

ordination, sets of neighboring cognitive radios self-organize themselves into coordi-

nation/bargaining groups and each such group modiﬁes spectrum assignment within

the group to improve system utility. The major challenge in designing a local coordi-

nation protocol is to ensure that the local improvements approach the global optimal

solution and the convergence speed is fast.

Two different approaches can be used for local coordination, namely, the explicit

bargaining-based approach and the implicit rule-based approach. In the former ap-

proach, the cognitive radios negotiate spectrum usage through message exchange,

while in the latter approach the radios observe the behaviors of neighboring nodes

and independently adjust spectrum usage following predeﬁned rules. The explicit

approach will incur overheads in terms of communication signaling and transmis-

sion energy. The implicit approach may be simpler to implement with signiﬁcant

reduction in signalling trafﬁc among the radios and power consumption.

To facilitate bargaining in the explicit bargaining-based approach, the authors

propose two constraints, namely, the limited neighbor bargaining constraint and the

self-contained group bargaining constraint. With the former constraint, bargaining

within a group is coordinated by a central leader, and it is limited among the leader

XX Preface

and its k-hop neighbors. With the latter constraint, bargaining inside each group

should not disturb the spectrum assignments in other groups. Therefore, a local im-

provement should lead to a system improvement. Under this constraint, the number

of channels which are exchanged among nodes inside a bargaining group is limited

and also the members of any two bargaining groups are not directly connected.

The authors propose a local bargaining protocol, namely, Fairness Bargaining

with Feed Poverty based on the explicit negotiation approach. In this protocol, a cog-

nitive radio willing to improve its spectrum usage starts bargaining with its neighbors

on a one-to-one basis to improve system utility. However, if there is no negotiable

channel found between it and any of its neighbors, the node initiates a Feed Poverty

Bargaining so that the neighboring nodes can collaborate together to feed it with

some channels. For this protocol, the lower bound on the throughput performance

(i.e., poverty line) for each of the nodes is obtained. Simulation results show that

compared to a centralized approach (e.g., based on a graph multi-coloring approach),

the proposed local bargaining approach incurs much lower communication overhead

while achieving similar performance (in terms of fairness utility). Note that, each

iteration of spectrum assignment/bargaining involves a four-way handshake (i.e., re-

quest, acknowledgment, action, acknowledgment) among neighbors. As expected,

the complexity of the bargaining approach increases with increase in the rate of

change of network topology (i.e., user mobility). System utility scales inversely with

increase in user density. With a ﬁxed user density, the system overhead scales linearly

with the number of users. Therefore, the local bargaining-based approach would be

suitable for large scale networks.

MAC Protocols for Hardware-Constrained Cognitive Wireless Networks

Chapter 13, authored by Qian Zhang, Juncheng Jia, and Xuemin Shen, presents a

single radio multi-channel MAC protocol for hardware-constrained cognitive wire-

less networks. The cognitive radios need to sense spectrum before transmission. The

MAC layer in a cognitive radio determines when and which channel it should sense

and then physical layer techniques (e.g., energy detection, matched ﬁlter detection,

etc.) are used to detect the primary users’ signal. The hardware constraints in a cog-

nitive radio are due to the sensing constraint and transmission constraint. Since at a

given time a practical cognitive radio may be able to sense only a small portion of

the radio spectrum, this gives rise to the sensing constraint. The transmission con-

straint arises due to the spectrum fragmentation (e.g., a cognitive radio may be able

to spread the transmission signal within a limited number of spectrum fragments).

Again, there is a constraint on the maximum amount of time a primary user can

tolerate interference from the secondary user (as in IEEE 802.22) – this is referred

to as the transmission parameter limitation. This parameter dictates how quickly a

cognitive radio must be able to detect incumbents. The problem is then to optimize

the sensing decision during each sensing and transmission interval. Note that, the

more spectrum is sensed, the more spectrum opportunity can be explored. The pro-

posed hardware-constrained MAC protocol takes the sensing constraint and sensing

overhead as well as the transmission parameter limitation into account.

Preface XXI

The authors model the sensing process as an optimal stopping problem with a ﬁ-

nite horizon which can be described as the problem of choosing a time to take an ac-

tion based on sequentially observed random variables in order to maximize/minimize

an expected payoff/cost. Such a problem can be solved by using the backward induc-

tion principle. However, it incurs exponential computational complexity. To reduce

the complexity for a practical MAC protocol, the authors use a k-stage look-ahead

rule at each stage to decide whether to stop or continue sensing.

In the considered system model, a common control channel is used for contention-

based random access (e.g., based on IEEE 802.11 DCF) by the cognitive radios to

reserve the time interval for spectrum sensing and subsequent data transmission be-

tween a cognitive radio transmitter and receiver. The time frame in HC-MAC is di-

vided into three parts: contention, sensing, and transmission. The contention part

is used to reserve the sensing period. During the sensing part, the cognitive radio

transmitter and the receiver sense the spectrum channels and determine whether a

channel is available at both the sides. The optimal stopping rule is used to decide

when to stop sensing. In the transmission part, actual data transmission takes place.

Simulation results show that, under different system conﬁgurations, the proposed

HC-MAC protocol achieves better system throughput compared to a static sensing

scheme where the cognitive radios always sense a ﬁxed number of channels.

Microeconomic Models for Dynamic Spectrum Management

Chapter 14, authored by Dusit Niyato and Ekram Hossain, investigates the problem

of dynamic spectrum sharing and pricing using

the oligopoly market models from

microeconomic theory. In an oligopoly market model, a number of producers (i.e.,

oligopolists) compete with each other independently to maximize their utility by

controlling the quantity and/or the price of the supplied commodity. The classical

oligopoly models include Cournot, Stackelberg, and Bertrand models which differ

from one another in terms of the nature of the competition. In the Cournot model,

producers compete in terms of amount of commodity to be supplied to the market.

In the Stackelberg model, some producers referred to as leaders are able to make

decisions on the amount of supplied quantity before other producers referred to as

followers. The followers make decision on the amount of supplied quantity by taking

the leader’s decision into account. In the Bertrand model, all producers make decision

simultaneously in terms of price. All of these models can be analyzed by using game

theory.

Oligopoly market models can be used for analyzing dynamic spectrum manage-

ment in cognitive radio networks. One of the objectives of spectrum management

here is to maximize the proﬁts of the cognitive radios (e.g., in terms of spectrum

share) and/or the primary service providers/users (e.g., in terms of spectrum price).

This is the same as the objective of an oligopoly competition. The oligopoly mar-

ket models were well studied in economics. Also, they are computationally simple,

and therefore, suitable for implementation in resource-limited software deﬁned radio

transceiver.

XXII Preface

The authors demonstrate the applications of Cournot, Stackelberg, and Bertrand

models of competition for spectrum/bandwidth sharing and pricing in cognitive wire-

less networks. Speciﬁcally, these three different models for oligopoly are applied to

obtain the optimal size of spectrum/bandwidth sharing and the charging price. The

Cournot game model is used for the case where multiple cognitive radios share the

spectrum/bandwidth with a primary user and the objective is to maximize the proﬁt

of the cognitive radios. Here, all the cognitive radios can completely observe the

strategies and the payoffs of other secondary users. The proﬁt of a cognitive radio is

a decreasing function of the amount of spectrum requested (and subsequently allo-

cated) to other users (i.e., their strategies). Again, the proﬁt is an increasing function

of the amount of spectrum allocation and the quality of the corresponding spectrum

opportunity. The Nash equilibrium is considered as the solution of the spectrum com-

petition which ensures that all of the cognitive radios are satisﬁed with the solution.

The Nash equilibrium is obtained by using the best response functions.

In the Bertrand model, several service providers (or primary users) compete with

each other in terms of price to gain the highest proﬁt under QoS constraints for

the primary users. Here, the bandwidth demand of the cognitive radios is established

based on a utility function which depends on the quality of transmission (i.e., channel

quality) in the available spectrum as well as the spectrum price charged by the pri-

mary user/service provider. The payoffs for the primary users are determined based

on the spectrum price and the spectrum demand from the secondary users and the

cost of sharing the spectrum with the secondary users. In addition, the authors con-

sider spectrum substitutability in the formulation which represents the ability of a

cognitive radio to switch among the frequency spectra offered by different primary

users. Again, the Nash equilibrium is obtained as the solution of this price competi-

tion.

The Stackelberg leader-follower competition is used to model the problem of

optimal sharing and pricing under elastic bandwidth demand from the cognitive ra-

dios. The authors illustrate the application of this competition model in the context

of spectrum sharing in an integrated WiMAX/WiFi network where the WiFi nodes

share the licensed WiMAX spectrum for broadband Internet access. The WiMAX

base stations (BSs) and the WiFi access points (APs) are operated by different ser-

vice providers. The WiMAX BS and the WiFi APs are the leader and the followers,

respectively. The Stackelberg equilibrium is considered as the solution of this spec-

trum sharing game.

Numerical performance evaluation results are presented for all of these oligopoly

competition models to show their efﬁcacy in allocating radio resource in cognitive

radio environments.

Analysis of Cognitive Radio Dynamics

Chapter 15, authored by Maria-Gabriella Di Benedetto et al., presents a mathematical

framework for analyzing the behavior of a self-organizing cognitive radio network

where continuous dynamics and discrete processes tightly interact.

This framework

剩余457页未读，继续阅读

mummyok

粉丝: 0
资源: 8

认知无线电通信网络：探索无线通信新边界

Cognitive Wireless Communication Networks（经典）

Cognitive Wireless Communication Networks

Cognitive wireless communication networks

告诉我关于基于LSTM的频谱预测这个题目的论文的近五年的英文参考文献

Intelligent Reflecting Surface-Aided Spectrum Sensing for Cognitive Radio

fuzzy cognitive是深度学习吗

c# Cognitive Services

cognitive complexity of method

能量检测算法和fcme算法的国内外研究现状

Cognitive Services 主要有哪些应用？可以用来作什么？

最新资源