WO2020181597A1 - 大规模mtc网络中一种基于分组的随机接入与数据传输方法 - Google Patents

大规模mtc网络中一种基于分组的随机接入与数据传输方法 Download PDF

Info

Publication number
WO2020181597A1
WO2020181597A1 PCT/CN2019/081475 CN2019081475W WO2020181597A1 WO 2020181597 A1 WO2020181597 A1 WO 2020181597A1 CN 2019081475 W CN2019081475 W CN 2019081475W WO 2020181597 A1 WO2020181597 A1 WO 2020181597A1
Authority
WO
WIPO (PCT)
Prior art keywords
gateway
link
mtcg
random access
successfully
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2019/081475
Other languages
English (en)
French (fr)
Inventor
王熠晨
杨子欢
王璐
王弢
李壮
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian Jiaotong University
Original Assignee
Xian Jiaotong University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xian Jiaotong University filed Critical Xian Jiaotong University
Publication of WO2020181597A1 publication Critical patent/WO2020181597A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W24/00Supervisory, monitoring or testing arrangements
    • H04W24/02Arrangements for optimising operational condition
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W4/00Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/70Services for machine-to-machine communication [M2M] or machine type communication [MTC]
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W72/00Local resource management
    • H04W72/04Wireless resource allocation
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04WWIRELESS COMMUNICATION NETWORKS
    • H04W74/00Wireless channel access
    • H04W74/08Non-scheduled access, e.g. ALOHA

Definitions

  • the invention belongs to the random access and data transmission technology in a large-scale MTC network, and specifically relates to the realization of a packet-based random access and data transmission scheme in the large-scale MTC network.
  • MTCD machine-type communication devices
  • Massive machine-type communication, mMTC has subsequently become one of the main communication scenarios for 5G.
  • mMTC is mainly deployed in cellular networks.
  • the traditional cellular network is designed for H2H, and mMTC will have a series of problems due to its own characteristics: First, the number of MTCDs is huge, and the number of preambles used for access is small. Massive MTCDs are initiated to the base station in a short time.
  • Access requests will cause more access collisions, leading to serious network congestion, making a large number of devices unable to transmit data in time; secondly, the total amount of communication data in mMTC is large but the data packets of a single device are small, so traditional The four-way handshake process will cause large signaling overhead; finally, most of the MTCD transmission rate is small, but the wireless resources are limited.
  • the direct data transmission from the MTCD to the base station will greatly reduce the resource utilization, and the MTCD itself is not suitable for transmission rate The high-speed transmission characteristics of 5G. Therefore, we urgently need to optimize and improve the existing cellular network based on the characteristics of the MTC itself.
  • Some solutions start from the ACB mechanism to improve and alleviate the network congestion problem; some solutions focus on the reasonable allocation and use of resources to improve resource utilization; some solutions group devices, introduce gateways, and design double-hop networks To improve system performance. Among them, most of the grouping schemes do not specify the resource usage of the communication link from the device to the gateway, but only point out the use of a certain communication method.
  • Some studies have introduced the concept of gateway forwarding threshold, and programming simulations show the impact of gateway forwarding threshold on system performance, but the concept of gateway forwarding threshold has not been further substituted into theoretical analysis for analytical deduction and theoretical analysis of gateway forwarding The impact of the threshold on system performance.
  • the purpose of the present invention is to overcome the above-mentioned shortcomings of the prior art and provide a packet-based random access and data transmission method in medium and large-scale MTC networks.
  • the design is based on packet Random access and data transmission schemes, combined with the dual-hop link resource allocation ratio and gateway forwarding threshold, analyze and optimize the system to further improve network performance.
  • the MTC gateway broadcasts ACB-like factors to the activated devices in the group, and the activated devices perform ACB-like detection.
  • the devices in the group that pass the ACB-like detection randomly select a channel among U 1 orthogonal channels and transmit the data packet to the corresponding MTC.
  • MTCG detects U 1 orthogonal channel. Only when a certain channel is selected by a device, MTCG can correctly detect the corresponding data packet, and even back-off the corresponding device for the data packet that fails the detection;
  • the gateway receives the successfully detected data packets according to the buffer margin. If the buffer margin is greater than or equal to the number of successfully detected data packets, then all the successfully detected data packets are successfully received by the gateway and stored in the buffer queue, otherwise the gateway is buffered according to The margin size is randomly selected from the successfully detected data packets and stored in the cache queue, so that the cache queue length reaches the optimal forwarding threshold ⁇ * ;
  • the MTCG whose buffer queue length reaches the optimal forwarding threshold ⁇ * initiates an access request to the BS, and accesses the base station through a four-message handshake process based on contention-based random access.
  • the successful MTCG goes to the next step, and the access fails.
  • MTCG initiates an access request again at the next random access opportunity;
  • the base station randomly selects the successfully accessed MTCG according to the number of orthogonal channels of link 2 and allocates a channel for the gateway to forward the data packet in the buffer. If U 2 is greater than or equal to the number of successfully accessed MTCG, then all the successfully accessed All MTCGs can successfully forward data packets, otherwise the base station randomly selects U 2 among these MTCGs and allocates channels for the corresponding gateway to forward data packets in link 2. The MTCG that has not obtained the link 2 channel will initiate again at the next random access opportunity Access request.
  • the throughput optimization problem is
  • Q M represents the total number of resource blocks used by the system to transmit data in a random access opportunity
  • ⁇ 2 represents the number of resource blocks required by the gateway to forward a data packet
  • ⁇ U represents the upper bound of the actual cache of the gateway
  • represents the stability of the system Distribution
  • P represents the transition probability matrix of the gateway cache occupancy
  • N PA represents the number of preamble codes.
  • the modified differential evolution algorithm is 1) using a penalty function to transform the original optimization problem into an unconstrained optimization problem
  • g( ⁇ , ⁇ ) represents the penalty function
  • the integer variable forwarding threshold ⁇ is regarded as a real variable, and the real variable ⁇ is rounded to the nearest integer before the objective function is calculated.
  • the evolutionary algorithm mutation operation is not closed. When the resource allocation ratio ⁇ and the forwarding threshold ⁇ obtained through the mutation operation exceed the value range, they are randomly generated within the value range to replace the variables outside the boundary.
  • the penalty function g ( ⁇ , ⁇ ) T ( ⁇ , ⁇ ) - ⁇ [max (0, f ( ⁇ , ⁇ )) + max (0, - ⁇ ⁇ )], where ⁇ is a penalty factor.
  • the packet-based random access and data transmission scheme proposed by the present invention is more suitable for the mMTC network. For the entire system, it can greatly improve the throughput, reduce the total service time of the system, alleviate network congestion, and improve resource utilization. For each MTCD , Can reduce the average access delay and increase the probability of successful access.
  • Figure 1 is a system model of the double-hop mMTC network in the packet-based random access and data transmission scheme of the present invention.
  • Figure 2 shows the resource allocation of link 1 and link 2 in the packet-based random access and data transmission scheme of the present invention.
  • Fig. 3 is the variation curve of the total service time with the number of MTCDs of each group in the system under the present invention and the traditional scheme.
  • Figure 4 is the variation curve of the total throughput with the number of MTCD per group of the system under the present invention and the traditional scheme.
  • Fig. 5 is the variation curve of the average throughput with the number of MTCD per group of the system under the present invention and the traditional scheme.
  • Fig. 6 is the variation curve of the successful access probability with the number of MTCD per group of the system under the present invention and the traditional scheme.
  • Fig. 7 is the variation curve of the average access delay of successful equipment with the number of MTCD per group of the system under the present invention and the traditional scheme.
  • N G MTC gateways N G MTC gateways
  • each MTCG manages N g MTCDs.
  • N D N G in the network.
  • N g MTCDs in other words, N D MTCDs are divided into N G MTCD groups according to their MTCG.
  • a random access opportunity (RAO) is recorded as T RAO , and the time axis is divided by T RAO as a unit.
  • RAO random access opportunity
  • the mMTC network architecture designed by the present invention is a double-hop network.
  • the communication from the MTCD to the BS includes two links, called link 1 and link 2, respectively.
  • MTCD transmits data packets to MTCG through link 1
  • MTCG transmits data packets aggregated in the buffer to the BS through link 2.
  • ⁇ -resource allocation ratio The resources of link 1 are equally divided into N G parts, which are respectively allocated to N G MTCD groups. Assuming that MTCD in link 1 needs ⁇ 1 RB to transmit a data packet to MTCG, then each MTCD group in link 1 contains
  • link 2 only the gateway whose buffer size reaches the forwarding threshold ⁇ and successfully connected can transmit the data packets aggregated in the buffer to the BS through link 2. Note that we stipulate that when the gateway forwards the data packet to the BS, the data in the buffer queue should be transferred once All ⁇ data packets are transmitted to the base station. Assuming that MTCG transfers a data packet to BS in link 2 requires ⁇ 2 RBs, then link 2 contains
  • link 1 it is assumed that each MTCD in each RAO has at most one data packet to be sent.
  • the communication process of link 1 includes the following steps:
  • MTCG receives data packets according to the buffer margin.
  • the forwarding threshold ⁇ can initiate an access request to the base station, that is, a four-message handshake process.
  • the buffer of the gateway in the double-hop mMTC network of the present invention can hold ⁇ data packets at most.
  • the process of the gateway receiving data packets consists of two parts. 1) First, MTCG detects U 1 orthogonal channel. Each channel may have 3 situations, which are selected by one device, not selected by the device, and selected by multiple devices. Only one channel is selected by one device.
  • MTCG can correctly detect the corresponding data packet, we also say that the device U 1 is selected successfully.
  • the gateway then cache the received data packet successfully detected margin based, in particular, if the margin is greater than the cache U 1 equal to the number of packets successfully selected, then all U 1 selected gateway packet is successfully received and stored successfully In the cache queue, otherwise, the gateway randomly selects the data packets successfully selected by U 1 according to the size of the cache margin and stores them in the cache queue, so that the length of the cache queue reaches the forwarding threshold ⁇ .
  • each MTCD group contains U 1 orthogonal channels, according to the principle of ACB mechanism, each orthogonal channel corresponds to a preamble.
  • the ACB-like factor takes a value as follows:
  • This type of ACB mechanism can control each MTCD group to have U 1 active devices in each RAO, thereby maximizing the number of successful devices selected by U 1 .
  • the probability of a channel being selected by a device as then:
  • each MTCD group contains U 1 orthogonal channels
  • the value of the number of data packets that an MTCG can successfully detect in a RAO is 0, 1, 2, ..., U 1 .
  • the communication process of link 2 includes the following steps:
  • the MTCG whose buffer queue length reaches the forwarding threshold ⁇ initiates an access request to the BS, and accesses the base station through a contention-based random access four-message handshake process, that is, MTCG randomly selects a preamble and sends it to the BS through the physical random access channel
  • the BS detects the preamble and sends back a random access response message to the MTCG.
  • the MTCG transmits information on the allocated uplink channel through the random access response message.
  • the BS detects the information and sends a contention resolution message to the successfully detected MTCG to notify the MTCG to receive Into success.
  • the MTCG access is successful, which is also called the success of the MTCG competition preamble.
  • the MTCG that successfully competes for the preamble proceeds to the next step, and the MTCG that fails the competition initiates an access request again in the next RAO.
  • the BS randomly selects the MTCG that successfully competes for the preamble according to the number of orthogonal channels of link 2 and allocates a channel for the gateway to forward the data packet in the buffer. Specifically, if U 2 is greater than or equal to the number of successful MTCGs in the preamble, all successfully connected MTCGs can successfully forward data packets. Otherwise, the BS randomly selects U 2 among these MTCGs and allocates channels for the corresponding gateway in the chain Data packets are forwarded in way 2.
  • N PA number of preamble codes. So the number of successful MTCGs in the competition preamble is:
  • the probability of being successfully allocated to the link 2 channel is:
  • the probability that the MTCG that reaches the forwarding threshold successfully forwards the data packet is:
  • U 1 ⁇ indicates that the number of orthogonal channels in each group of link 1 is greater than or equal to the gateway forwarding threshold.
  • the transition probability matrix of the buffer size state is shown in equation (12).
  • P is a matrix of ⁇ +1 rows and ⁇ +1 columns.
  • U 1 ⁇ indicates that the number of orthogonal channels in each group of link 1 is less than the gateway forwarding threshold.
  • P G, k 0, (k>U 1 ), that is, in a RAO, a gateway is on the link 1
  • the number of successfully detected data packets cannot exceed U 1 .
  • the gateway Since a gateway can successfully detect U 1 data packets at most, when U 1 ⁇ , the gateway with a buffer size of at least ⁇ -U 1 may reach the forwarding threshold through the reception of link 1.
  • we can get the probability that the current RAO gateway will reach the forwarding threshold due to the successful reception of the data packet when U 1 ⁇ is shown in the formula (16) in the brackets on the right side of the equal sign.
  • the throughput of a RAO in the steady state of the system can be expressed as:
  • U 2 ⁇ N G,PA that is, the number of successfully connected gateways is greater than or equal to the number of channels of link 2. Therefore, the probability that a gateway that successfully competes for the preamble can be allocated to the channel of link 2 is:
  • U 2 > N G, PA that is, the number of gateways that have successfully accessed is less than the number of channels of link 2. All gateways that successfully compete for the preamble can be allocated to the channel of link 2, so the competition preamble The probability that a successful gateway can be assigned to the link 2 channel is:
  • Equation (24b) is the constraint on the value range of the resource allocation ratio.
  • Equation (24c) shows that the forwarding threshold is a positive integer with a minimum value of 1, but it cannot exceed the actual gateway cache upper bound ⁇ U.
  • Equations (24d) and (24e) are the constraints of the Markov chain stable distribution theory, which are used to solve the smooth distribution of gateway cache under the conditions of a given resource allocation ratio and forwarding threshold.
  • Equation (24d) shows the smooth distribution and transfer The relationship of the probability matrix, formula (24e) shows that the sum of the probabilities of all states under a stationary distribution is 1, and the probability of each state is greater than 0 and less than 1.
  • Equation (24f) indicates that the maximum throughput occurs in the first case of 5.1, and the corresponding optimization target expression is shown in Equation (24a).
  • P T -I is a matrix with ⁇ +1 rows and ⁇ +1 columns, and no matter U 1 ⁇ or U 1 ⁇ , the second row of P T -I
  • formula (44) is a quadratic equation of ⁇ ⁇ -1 . From the root formula, we can get:
  • the optimization problem described by equation (24) is a mixed integer nonlinear programming problem with constraints. Since the problem contains both continuous and discrete variables, it is difficult to solve. Therefore, we use a modified differential evolution algorithm to solve it.
  • Differential evolution algorithm is a heuristic algorithm based on real number coding and global optimization in continuous space. It intelligently guides the direction of optimization search through the group generated by the cooperation and competition of individuals between groups, including population initialization, mutation, Cross, choose 4 operations. Next, we will introduce the specific content of these four operations in differential evolution.
  • the specific operation of population initialization is to randomly and uniformly generate the first generation population according to the value range of each component:
  • the mutant individuals produced by the mutation operation are:
  • s 1 , s 2 , s 3 3 individuals randomly selected in the population and s 1 ⁇ s 2 ⁇ s 3 ⁇ m, F——variation factor, generally F ⁇ (0,2).
  • Crossover operation refers to dividing the parent individual according to a certain probability And variant individuals Cross each component separately to generate a temporary individual.
  • Each component of the temporary individual generated by the mth individual of the lth generation population through this operation is:
  • nth component of, r 1,n (n 1, 2)—— Randomly uniform in the interval [0,1] Parts.
  • the selection operation refers to the parent individual With temporary individuals Choose the best individual as the next generation individual. For the maximization problem, it is to choose the next generation that makes the objective function larger, and the parent individual
  • the next-generation individuals produced by this operation are:
  • g() represents the objective function in the optimization problem.
  • the mutation individual may have boundary violations, that is, the value of a component may exceed the boundary. Therefore, when the value of a component exceeds the boundary, we take it A number is randomly generated within the value range to replace the component.
  • Figure 3 shows the variation curve of the total service time with the number of MTCDs in each group under the two schemes.
  • the total service time is defined as the total number of RAOs required for the system to service all MTCDs. All devices are fully serviced. The device either successfully transmits within the maximum number of transmissions, or fails due to exceeding the maximum number of retransmissions. It can be seen that under the two schemes, the total service time increases as the number of MTCDs in each group increases, but the grouping scheme in this paper has a slower growth rate, and the scheme designed by the present invention can greatly reduce the total service time compared to the traditional scheme.
  • the gateway is introduced into the traditional solution, and the double-hop mMTC network architecture is designed,
  • the Markov chain is used to analyze the smooth distribution of the gateway cache state, and on this basis, the system is optimized to maximize the system throughput, thereby improving the system performance and greatly reducing the total service time.
  • Figure 4 shows the variation curve of the total throughput with the number of MTCDs in each group of the system under the two schemes.
  • the total throughput is defined as the total number of MTCDs successfully transmitted within the maximum number of transmissions.
  • the total service time is shown in the figure.
  • the total throughput and the total throughput during the interception period It can be seen that under the two schemes, the total throughput in the total service time is equivalent. In other words, under the two schemes, almost all devices can transmit successfully within the maximum number of retransmissions. However, it can be seen from Figure 3 that the traditional scheme requires Spend more time.
  • the total throughput in the interception time period differs greatly between the two schemes. As the number of MTCDs in each group increases, the solution of the present invention can serve more devices in the interception time period. The specific reason is the same as that shown in FIG. 3.
  • Figure 5 shows the variation curve of the average throughput with the number of MTCDs in each group of the system under the two schemes.
  • the average throughput is defined as the number of MTCDs that can be served by an average RAO, that is, the number of MTCDs successfully accessed within the maximum number of transmissions.
  • the ratio of the total number of RAOs required for the service, the average throughput during the total service time and the average throughput during the interception period are plotted in the figure. It can be seen that under the two schemes, the average throughput in the total service time and the interception period is equivalent, but the scheme of the present invention can greatly improve the average throughput compared with the traditional scheme. The specific reason is similar to Figure 3.
  • Figure 6 shows the variation curve of the probability of successful access with the number of MTCDs in each group of the system under the two schemes.
  • the probability of successful access within the total service time and the interception time period is drawn respectively.
  • the probability of successful access It is defined as the ratio of the number of successfully accessed MTCDs to the total number of MTCDs transmitted in the defined time period. It can be seen that under the two schemes, the probability of successful access within the total service time period is equivalent, but the scheme of the present invention has a higher probability of successful access during the interception time period. The specific reason is similar to Figure 3.
  • Figure 7 shows the variation curve of the average access delay of successful devices with the number of MTCDs per group under the two schemes.
  • the average access delay is defined as the average RAO required for a successful MTCD from the initial activation to the successful reception of the base station. number. It can be seen that under the two schemes, during the total service time, the scheme of the present invention can greatly reduce the average delay. During the interception time period, the scheme of the present invention also reduces the average access delay of successful devices. The specific reason is similar to Figure 3.
  • the packet-based random access and data transmission scheme proposed by the present invention is more suitable for the mMTC network.
  • the throughput can be greatly improved, and the total service time of the system can be reduced.
  • it can reduce The average access delay increases the probability of successful access.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

本发明公开了大规模MTC网络中一种基于分组的随机接入与数据传输方法,在mMTC网络中,数据传输的资源分为链路1与链路2,MTCD通过链路1将数据包发送给网关,网关根据缓存余量接收数据包,缓存达到阈值的网关采用基于竞争的四步握手过程竞争接入基站,成功接入基站的网关将聚合的数据包通过链路2在分配的上行信道上以较高的速率传输给基站。推导出系统稳态时发起接入请求的网关数量,分析得到系统吞吐量的表达式。以最大化吞吐量为目标,结合资源分配比及网关转发阈值等约束条件构造了吞吐量优化问题,差分演化算法求出最优的资源分配比及网关转发阈值。本发明有效地提高系统吞吐量,降低接入时延。

Description

大规模MTC网络中一种基于分组的随机接入与数据传输方法 技术领域
本发明属于大规模MTC网络中的随机接入与数据传输技术,具体涉及大规模MTC网络中一种基于分组的随机接入与数据传输方案的实现。
背景技术
作为开启智能时代的关键环节,物联网已经逐步深入到社会各个领域,机器类通信设备(machine-type communication device,MTCD)的数量也在不断增长,大规模机器类通信(massive machine-type communication,mMTC)随之成为了5G的主要通信场景之一。为了便于广泛实施机器类通信,mMTC主要部署在蜂窝网络中。然而,传统蜂窝网络是为H2H设计的,mMTC由于其自身的特点会产生一系列问题:首先,MTCD数量庞大,而用于接入的前导码数量较少,海量MTCD在较短时间内向基站发起接入请求,会造成较多的接入碰撞,从而导致严重的网络拥塞,使得大量设备不能及时传输数据;其次,mMTC中通信数据总量较大但单个设备的数据包较小,采用传统的四步握手过程会造成较大的信令开销;最后,大多MTCD的传输速率较小,而无线资源有限,由MTCD直接向基站传输数据会大大降低资源利用率,且MTCD本身传输速率并不适用于5G的高速传输特性。因此,我们急需从MTC本身特点出发,对现有蜂窝网络进行优化与改进。一些方案从ACB机制出发做以改进,缓解了网络拥塞问题;一些方案针对资源的合理分配及使用加以研究,提升了资源利用率;还有一些方案对设备进行分组,引入网关,设计双跳网络来提升系统性能。其中,大多数分组方案并没有详细的说明由设备到网关这一通信链路的资源使用情况,只是指出采用某种通信方式。一些研究引入了网关转发阈值这一概念,通过编程仿真表明网关转发阈值对系统性能的影响,但是没有进一步将网关转发阈值这一概念代入到理论分析中,进行解析推导,从理论上分析网关转发阈值对系统性能的影响。
发明内容
本发明的目的在于克服上述现有技术不足,提供中大规模MTC网络中一种基于分组的随机接入与数据传输方法,考虑到网关转发阈值对于分组双跳mMTC网络系统的影响,设计基于分组的随机接入与数据传输方案,并联合考虑双跳链路的资源分配比及网关转发阈值,对系统进行分析与优化,从而进一步提升网络性能。
本发明的技术方案是这样解决的:包括以下步骤:
(1)在mMTC网络中,引入MTC网关MTCG,让每个网关管理若干个设备,分别对应于一个MTCD组,将用于数据传输的资源按照修正差分演化算法所求的最优资源分配比β *分为2部分:即链路1与链路2,将链路1中的资源均分给每组,每组的资源按照MTCD传输一个数据包所需的资源块数划分为U 1个信道,根据修正差分演化算法所求的最优转发阈值γ *及网关转发一个数据包所需的资源块数将链路2划分为U 2个信道;
(2)MTC网关向组内激活设备广播类ACB因子,激活设备进行类ACB检测,通过类ACB检测的组内设备在U 1个正交信道中随机选择一个信道将数据包传输给相应的MTC网关;
(3)MTCG检测U 1个正交信道,其中只有某个信道被一个设备选择时,MTCG才能正确检测出对应的数据包,对于检测失败的数据包其对应设备进行均匀退避;
(4)网关根据缓存余量接收成功检测的数据包,若缓存余量大于等于检测成功的数据包数,则所有检测成功的数据包被网关成功接收并存储在缓存队列中,否则网关根据缓存余量大小在检测成功的数据包中随机选择若干个并存储在缓存队列中,使得缓存队列长度达到最优转发阈值γ *
(5)缓存队列长度达到最优转发阈值γ *的MTCG向BS发起接入请求,通过基于竞争的随机接入四消息握手过程接入基站,接入成功的MTCG进行下一步,接入失败的MTCG在下一个随机接入机会再次发起接入请求;
(6)基站根据链路2的正交信道数随机选择接入成功的MTCG并分配信道供网关转发缓存中的数据包,若U 2大于等于接入成功的MTCG数,则所有接入成功的MTCG都可以成功转发数据包, 否则基站在这些MTCG中随机选择U 2个并分配信道供相应网关在链路2中转发数据包,未获得链路2信道的MTCG在下一个随机接入机会再次发起接入请求。
所述步骤(1)中最优资源分配比β *和最优转发阈值γ *,首先通过马尔可夫理论分析出网关缓存队列长度的状态转移概率矩阵,对网关缓存占有量的变化过程进行数学建模,得到每一个随机接入机会中发起接入请求的网关数,即
Figure PCTCN2019081475-appb-000001
π i,(i=0,1,...γ)表示系统达到平稳状态时网关缓存大小在每一RAO结束时刻为i的概率,N G表示网关数,P G,t(i=0,1,2,...,U 1)表示一个MTCG在一个RAO成功检测i(i=0,1,2,...,U 1)个包的概率,从而得到每一个随机接入机会的吞吐量,即T(β,γ)=N G,γ·P s·γ,P s表示网关成功转发的概率,结合资源分配比网关缓存上界以及平稳分布的条件构造吞吐量的优化问题,利用修正差分演化算法求得最优转发阈值γ *和最优资源分配比β *
所述的吞吐量的优化问题如下:
Figure PCTCN2019081475-appb-000002
s.t.0<β<1                            (24b)
γ∈N +,1≤γ≤γ U                      (24c)
Π T=Π TP                          (24d)
Figure PCTCN2019081475-appb-000003
Figure PCTCN2019081475-appb-000004
Q M表示系统在一个随机接入机会中用于传输数据的总资源块数,δ 2表示网关转发一个数据包所需的资源块数,γ U表示网关实际缓存上界,Π表示系统的平稳分布,P表示网关缓存占有量的转移概率矩阵,N PA表示前导码数。
所述的修正差分演化算法是1)利用罚函数将原优化问题转化为无约束优化问题,
Figure PCTCN2019081475-appb-000005
s.t.0<β<1                (56b)
γ∈N +,1≤γ≤γ U           (56c)
Π=ΠP                (56d)
Figure PCTCN2019081475-appb-000006
其中,g(β,γ)表示罚函数;2)在算法迭代过程中,将整型变量转发阈值γ看作实型变量再求目标函数之前将实型变量γ四舍五入取整,3)由于差分演化算法变异操作不封闭,当通过变异操作得到的资源分配比β和转发阈值γ超出取值范围时,在取值范围内随机生成以取代超出边界的变量。
所述罚函数g(β,γ)=T(β,γ)-σ[max(0,f(β,γ))+max(0,-π γ)],其中σ为惩罚因子。
本发明提出的基于分组的随机接入与数据传输方案更适用于mMTC网络,对于整个系统,可以大大提升吞吐量,减小系统总服务时间,缓解网络拥塞,提升资源利用率,对于每个MTCD,可以减小平均接入时延,增加成功接入概率。
附图说明
图1是本发明基于分组的随机接入与数据传输方案中双跳mMTC网络的系统模型。
图2是本发明基于分组的随机接入与数据传输方案中的链路1与链路2的资源分配情况。
图3是本发明与传统方案下的总服务时间随系统中每组MTCD数的变化曲线。
图4是本发明与传统方案下的总吞吐量随系统每组MTCD数的变化曲线。
图5是本发明与传统方案下的平均吞吐量随系统每组MTCD数的变化曲线。
图6是本发明与传统方案下的成功接入概率随系统每组MTCD数的变化曲线。
图7是本发明与传统方案下的成功设备平均接入时延随系统每组MTCD数的变化曲线。
具体实施方式
1.系统模型:
在mMTC网络中,有一个基站(base station,BS),其覆盖范围内包含N G个MTC网关(MTC gateway,MTCG),每个MTCG管理N g个MTCD,网络中共有N D=N G·N g个MTCD,换而言之,N D个MTCD按照所属MTCG被分为N G个MTCD组。一个随机接入机会时长(random access opportunity,RAO)记为T RAO,将时间轴以T RAO为单位进行划分,我们假设所有MTCD的流量到达模型服从时长为T a个RAO的均匀分布。在本发明中,我们假定所有的信道增益都是独立的,平稳的,并且各态历经的块衰落过程。
本发明设计的mMTC网络架构是一个双跳网络。如图1所示,由MTCD到BS的通信包含2条链路,分别称之为链路1与链路2。其中,MTCD通过链路1将数据包传给MTCG,MTCG通过链路2将缓存中聚合的数据包传给BS。
链路1与链路2的资源分配如图2所示。假设一个RAO用于传输数据的资源块(resource block,RB)共有Q M个,链路1所含资源块总数为
Q 1=βQ M,          (1)
其中:β——资源分配比。将链路1的资源均分为N G个部分,分别分配到N G个MTCD组中。假设链路1中MTCD向MTCG传输一个数据包需要δ 1个RB,则链路1中每个MTCD组含有
Figure PCTCN2019081475-appb-000007
个正交信道。
链路2中,只有缓存大小达到转发阈值γ且接入成功的网关才能通过链路2向BS传输缓存中聚合的数据包,注意我们规定网关向BS转发数据包时,一次把缓存队列中的γ个数据包全部传输给基 站。假设链路2中MTCG向BS中转一个数据包需要δ 2个RB,则链路2含有
Figure PCTCN2019081475-appb-000008
个正交信道供网关转发聚合的数据包。
2.链路1与链路2的通信方式
2.1链路1
链路1中,假设每个RAO中每个MTCD至多有一个数据包需要发送。链路1的通信过程包含以下步骤:
(1)MTCG向组内设备广播类ACB因子。每个网关估计该组当前RAO发送数据包的MTCD数N a,t(i=1,...N G),即该组在当前RAO中的激活设备数,本发明中假设每个网关都能精确估计N a,t。然后每个网关向该组所有设备广播类似于ACB机制的类ACB因子ρ来控制组内向网关传输数据的设备数,本发明中我们称该机制为类ACB机制。
(2)激活设备进行类ACB检测。具体来说,该组所有激活设备随机生成0~1之间的一个随机数,并与类ACB因子ρ进行比较,如果生成的随机数小于等于ρ,则该设备通过类ACB检测,否则进行均匀退避。对于每个MTCD组,通过类ACB检测的组内设备在U 1个正交信道中随机选择一个信道,并通过该信道将数据包传输给MTCG,如图1所示。
(3)MTCG根据缓存余量接收数据包。注意本发明中只有网关达到转发阈值γ才能向基站发起接入请求,即进行四消息握手过程,虽然网关实际缓存上界γ U≥γ,但是为了后续便于分析网关缓存状态的平稳分布,我们规定本发明的双跳mMTC网络中网关的缓存最多能容纳γ个数据包。网关接收数据包的过程包含2部分。1)首先,MTCG检测U 1个正交信道,每个信道可能出现3种情况,分别是被一个设备选择,没有被设备选择以及被多个设备选择,其中只有某个信道被一个设备选择时,MTCG才能正确检测出对应的数据包,我们也称该设备U 1选择成功。2)然后,网关根据缓存余量接收成功检测的数据包,具体来说,若缓存余量大于等于U 1选择成功的数据包数,则所有U 1选择成功的数据包被网关成功接收并存储在缓存队列中,否则网关根据缓存余量大小在U 1选择成功的数据包中进行随机选择并存储在缓存队列中,使得缓存队列长度达到转发阈值γ。
由于每个RAO中,每个MTCD组含有U 1个正交信道,根据ACB机制原理,每个正交信道相应于一个前导码,为了最大化U 1选择成功的设备数,类ACB因子取值如下:
ρ=min{U 1/N a,i,1}           (4)
鉴于考虑的是mMTC场景,每个RAO中每组需要传输数据的设备数较多,所以
ρ=U 1/N a,i            (5)
该类ACB机制可以控制每个MTCD组在每个RAO中有U 1个激活设备,从而最大化U 1选择成功的设备数。记一个信道被一个设备选择的概率为
Figure PCTCN2019081475-appb-000009
则:
Figure PCTCN2019081475-appb-000010
由于每个MTCD组含有U 1个正交信道,因此一个MTCG在一个RAO中能成功检测的数据包数的取值为0,1,2,…,U 1。记一个MTCG在一个RAO成功检测i(i=0,1,2,...,U 1)个包的概率为P G,t(i=0,1,2,...,U 1),则:
Figure PCTCN2019081475-appb-000011
2.2链路2
链路2的通信过程包含以下步骤:
(1)缓存队列长度达到转发阈值γ的MTCG向BS发起接入请求,通过基于竞争的随机接入四消息握手过程接入基站,即MTCG随机选择一个前导码通过物理随机接入信道发送给BS,BS检测前导码并向MTCG回发随机接入响应消息,MTCG通过随机接入响应消息在分配的上行信道传输信息,BS检测信息并对成功检测的MTCG发送竞争解决消息,以通知该MTCG接入成功。本发明中,我们规定若某一前导码仅被一个MTCG选择,则该MTCG接入成功,也称该MTCG竞争前导码成功。竞争前导码成功的MTCG进行下一步,竞争失败的MTCG在下一个RAO再次发起接入请求。
(2)BS根据链路2的正交信道数随机选择竞争前导码成功的MTCG并分配信道供网关转发缓存中的数据包。具体来说,若U 2大于等于竞争前导码成功的MTCG数,则所有接入成功的MTCG都可以成功转发数据包,否则BS在这些MTCG中随机选择U 2个并分配信道供相应网关在链路2中转发数据包。
令系统稳态时由于达到转发阈值而竞争前导码的MTCG数为N G,γ,则达到阈值的网关竞争前导码成功的概率为:
Figure PCTCN2019081475-appb-000012
其中,N PA——前导码数。所以竞争前导码成功的MTCG数为:
N G,PA=N G,γ·P PA              (9)
在这些竞争前导码成功的MTCG中,能被成功分配到链路2信道的概率为:
Figure PCTCN2019081475-appb-000013
因此,达到转发阈值的MTCG成功转发数据包的概率为:
Figure PCTCN2019081475-appb-000014
3.应用马尔科夫理论分析MTCG缓存状态的平稳分布
为了求解所设计的mMTC网络系统达到平稳状态时在每个RAO中缓存达到阈值的MTCG数N G,γ,本小节我们以网关缓存队列长度为状态,应用马尔科夫理论分析缓存大小的转移概率矩阵,从而推导N G,γ。每个网关的缓存大小状态共有γ+1种,分别为:0,1,2,…γ。分析缓存大小的概率转移矩阵需从以下2种情况讨论:
(1)U 1≥γ时
U 1≥γ表明链路1每组的正交信道数大于等于网关转发阈值,这种情况下,缓存大小状态的转移概率矩阵如式(12)所示。其中,P为γ+1行γ+1列的矩阵,为了便于后续推导,我们分别记为第0行,第1行,…第γ行,第0列,第1列,…第γ列,P ij(i=0,1,2,...γ,j=0,1,2,...γ)——表示缓存大小由i转移到j的概率。当网关在上一RAO结束时缓存大小为i(i=0,1,2,…γ-2),当前RAO在链路一的U 1个正交信道中成功检测j-i个数据包(j=i+1,i+2,...γ-1),且由于缓存余量大于成功检测的包数从而成功接收所有数据包使得缓存大小变为j时,由式(7),有P ij=P G,j-1,(i=0,1,2,...γ-2,j=i+1,...γ-1)。当网关在上一RAO结束时缓存大小为i(i=1,2,…γ-1),当前RAO在链路一的U 1个正交信道中成功检测0个数据包时,缓存大小保持不变,有P tt=P G,0,(i=1,2,...γ-1)。由于网关中的缓存只有达到阈值时才能有成功转发的机会且一次性 转发γ个数据包从而使得缓存变为0,所以有P ij=0,(i=2,...γ,j=1,2,...i-1)。当网关在上一RAO结束时缓存大小为i(i=0,2,…γ),当前RAO在链路一的U 1个正交信道中成功检测的数据包大于等于γ-i时,网关根据缓存余量大小对成功检测的数据包进行随机选择并接收到缓存队列中,使得缓存队列长度达到阈值,这种情况下,由式(11),若网关成功转发缓存中的数据包,则缓存大小清零,所以有
Figure PCTCN2019081475-appb-000015
由于缓存上一RAO为0,当前RAO没有成功检测数据包时,缓存仍为0,所以有
Figure PCTCN2019081475-appb-000016
若网关转发失败,则缓存大小保持为γ,所以有
Figure PCTCN2019081475-appb-000017
Figure PCTCN2019081475-appb-000018
记系统达到平稳状态时网关缓存大小在每一RAO结束时刻为i的概率为π i,(i=0,1,...γ),则系统的平稳分布为:
Π=[π 012,,...π γ] T           (13)
由此可以得到U 1≥γ时,系统在平稳状态下每RAO由于达到转发阈值而发起接入请求的网关数为:
Figure PCTCN2019081475-appb-000019
式(14)等号右边中括号里的式子表示当前RAO网关由于在链路1成功接收数据包使得缓存大小达到阈值的概率,由于U 1≥γ,所以上一RAO无论缓存大小为多少,当前RAO网关通过链路1数据包的接收都有可能达到转发阈值。也就是说,上一RAO结束时网关缓存大小为j(j=0,1,2,...γ)时,若网关当前RAO能成功检测的数据包数大于等于γ-j,则该网关能够达到转发阈值,因此该网关 当前RAO能够达到转发阈值的概率为
Figure PCTCN2019081475-appb-000020
(2)U 1<γ时
U 1<γ表明链路1每组的正交信道数小于网关转发阈值,这种情况下,定义P G,k=0,(k>U 1),即在一个RAO内一个网关在链路1成功检测的数据包数不可能超过U 1。根据U 1≥γ情况的讨论,同理我们有P ij(i=0,1,2,...γ-2,j=i+1,...γ-1)=P G,j-1,P tt(i=1,2,...γ-1)=P G,0,P ij(i=2,...γ,j=1,2,...i-1)=0。由于链路1每组的正交信道数小于网关转发阈值,所以缓存大于等于γ-U 1时才有转移到0或γ的可能,因此有
P 00=P G,0,P 0j=0,(j=1,..,γ-U 1-1),
P =0(i=0,1,...,γ-U 1-1),
Figure PCTCN2019081475-appb-000021
基于以上讨论,U 1<γ时缓存大小状态的转移概率矩阵如式(15)所示。
Figure PCTCN2019081475-appb-000022
同理可以得到U 1<γ时,系统在平稳状态下每RAO由于达到转发阈值而发起接入请求的网关数为:
Figure PCTCN2019081475-appb-000023
由于一个网关最多能成功检测U 1个数据包,所以U 1<γ时缓存大小至少为γ-U 1的网关才有可能通过链路1的接收达到转发阈值。也就是说,上一RAO结束时网关缓存大小为j(j=γ-U 1,...,γ)时, 若网关当前RAO能成功检测的数据包数大于等于γ-j,则该网关能够达到转发阈值,因此该网关当前RAO能够达到转发阈值的概率为
Figure PCTCN2019081475-appb-000024
基于以上分析,我们可以得到U 1<γ时当前RAO网关由于成功接收数据包而达到转发阈值的概率如式(16)等号右边中括号里的式子所示。
4.吞吐量及优化问题的描述与建立
在本小节中,我们首先分析系统稳态时吞吐量的表达式,然后结合平稳分布,转发阈值,资源分配比等约束条件,构建吞吐量优化问题。
4.1吞吐量分析
系统稳态时,BS成功接收的数据包数为成功转发的网关数与转发阈值的乘积,因此系统稳态时一个RAO的吞吐量可以表示为:
T(β,γ)=N G,γ·P s·γ           (17)
根据式(11),我们需要从以下2种情况进一步分析吞吐量的表达式:
Figure PCTCN2019081475-appb-000025
这种情况下,U 2≤N G,PA,即接入成功的网关数大于等于链路2的信道数,因此竞争前导码成功的网关能够分配到链路2信道的概率为:
Figure PCTCN2019081475-appb-000026
所以,达到阈值的网关能够成功转发的概率为:
Figure PCTCN2019081475-appb-000027
将式(19)代入式(17),可以得到
Figure PCTCN2019081475-appb-000028
时系统稳态下的吞吐量为:
Figure PCTCN2019081475-appb-000029
Figure PCTCN2019081475-appb-000030
这种情况下,U 2>N G,PA,即接入成功的网关数小于链路2的信道数,所有竞争前导码成功的网关都能被分配到链路2的信道,因此竞争前导码成功的网关能够分配到链路2信道的概率为:
Figure PCTCN2019081475-appb-000031
所以,达到阈值的网关能够成功转发的概率为:
Figure PCTCN2019081475-appb-000032
将式(22)代入式(17),可以得到
Figure PCTCN2019081475-appb-000033
时系统稳态下的吞吐量为:
Figure PCTCN2019081475-appb-000034
4.2优化问题的描述与建立
基于5.1的分析,当
Figure PCTCN2019081475-appb-000035
时,链路2的信道数大于接入成功的网关数,系统中链路2的一些资源未使用,表明该情况下,链路2分配的资源较多。如果把链路2中分配过多的没被使用的资源取出一部分划分到链路1中,则可以提升链路1中一个信道成功接收一个包的概率,从而增加网关缓存大小处于阈值的概率,适当增加成功竞争PA的网关数,提升吞吐量;再或者如果增加转发阈值,使链路2资源尽量被使用,也可以提升吞吐量。根据以上讨论,我们可以得出系统吞吐量最大值出现在第一种情况中。为了优化所设计的mMTC网络性能,我么通过优化资源分配比及网关转发阈值,来最大化系统稳态时的吞吐量,因此我们的吞吐量优化问题可以表示为:
Figure PCTCN2019081475-appb-000036
s.t.0<β<1                             (24b)
γ∈N +,1≤γ≤γ U                       (24c)
Π T=Π TP                            (24d)
Figure PCTCN2019081475-appb-000037
Figure PCTCN2019081475-appb-000038
其中,
Figure PCTCN2019081475-appb-000039
由式(14)与(16),有:
Figure PCTCN2019081475-appb-000040
式(24b)为资源分配比的取值范围约束,式(24c)表明转发阈值取值为正整数,最小值为1,但又不能超过实际中网关的缓存上界γ U。式(24d)及式(24e)为马尔科夫链平稳分布理论本身的约束,用于求解给定资源分配比及转发阈值条件下网关缓存的平稳分布,其中式(24d)表明平稳分布与转移概率矩阵的关系,式(24e)表明平稳分布下所有状态的概率之和为1,且每个状态的概率大于0小于1。式(24f)表示吞吐量最大值出现在5.1的第一种情况中,相应的优化目标的表达式如式(24a)所示。
5.优化问题的求解
本发明中的优化问题直接求解较为困难,很难得到闭式解,所以我们采用修正的差分演化算法来求解最优解。在差分演化算法的每次迭代中,我们首先需要根据式(24d)及(24e)求解缓存状态的平稳分布,根据式(12)和(15)所表示的概率转移矩阵,我们可以看出平稳分布也很难直接求出闭式表达式,因此,我们通过高斯约旦消元法求解平稳分布。所以本小节包含2部分内容:应 用高斯约旦消元法求解平稳分布以及应用改进的差分演化算法求解最优解。
5.1应用高斯约旦消元法求解平稳分布
本节中,我们首先根据式(24d)及(24e)求解缓存状态的平稳分布,从而进一步求解系统稳态时达到转发阈值的网关数N G,γ
由式(24d),我们有:
(P T-I)Π=0           (26)
式(26)为一个齐次方程组,包含了γ+1个方程,分别对应P T-I的γ+1行,其中,P T-I需要分以下2种情况讨论:
(1)U 1≥γ时
P T-I如式(27)所示:
Figure PCTCN2019081475-appb-000041
(2)U 1<γ时
P T-I如式(28)所示:
Figure PCTCN2019081475-appb-000042
根据式(27)和(28),我们可以看出P T-I为γ+1行γ+1列的矩阵,且无论U 1≥γ还是U 1<γ,P T-I的第2行到第γ行的第γ+1列元素均为0,也就是说,我们可以先根据这些行所对应的方程 组求出π i(i=0,1,2...,γ-2)与π γ-1的关系,然后将所求结果代入式(26)的第一个方程,求出π γ-1与π γ的关系,最后再把上述结果代入式(24e),我们就可以得到系统稳态时网关缓存大小的平稳分布。具体来说,我们首先把P T-I的第2行到第γ行的前γ列元素按照相应位置提取出来,构成矩阵A (γ-1)×γ,记Π 1=[π 01,...π γ-1] T,则可以得到齐次方程组:
1=0              (29)
利用高斯约旦消元法将系数矩阵A化为简化行阶梯式矩阵A 1,我们可以得到:
Figure PCTCN2019081475-appb-000043
其中,a i(i=0,1,...γ-2)为β,γ的函数。又因为
A 1Π 1=0                (31)
所以有:
π i+a iπ γ-1=0,(i=0,1,...γ-2)         (32)
因此,我们可以得到:
π i=-a iπ γ-1,(i=0,1,...γ-2)(33)
令a γ-1=-1,则有:
π i=-a iπ γ-1,(i=0,1,...γ-1)        (34)
将式(34)代入式(25),可以得到:
Figure PCTCN2019081475-appb-000044
为了进一步推导平稳分布及N G,γ,需从以下2种情况讨论:
(1)U 1≥γ时:
由式(27)的第一行对应的方程组有:
Figure PCTCN2019081475-appb-000045
整理可得
Figure PCTCN2019081475-appb-000046
将式(34)代入式(37),可以得到:
Figure PCTCN2019081475-appb-000047
(2)U 1<γ时:
由式(28)的第一行对应的方程有:
Figure PCTCN2019081475-appb-000048
将式(34)代入式(38),可以得到:
Figure PCTCN2019081475-appb-000049
基于以上2种情况的讨论,由式(38)及(40),我们可以得到无论U 1≥γ还是U 1<γ,均有下式成立:
Figure PCTCN2019081475-appb-000050
将式(35)代入(19),可以得到:
Figure PCTCN2019081475-appb-000051
将式(42)代入式(41)可以得到:
Figure PCTCN2019081475-appb-000052
整理可得:
Figure PCTCN2019081475-appb-000053
可以看出式(44)是π γ-1的二次方程,由求根公式,我们可以得到:
Figure PCTCN2019081475-appb-000054
由式(24e)及式(34),我们可以知道a j<0,(j=0,1,2,...γ-1),考虑到约束条件式(24e),我们可以得到:
Figure PCTCN2019081475-appb-000055
将式(34)代入式(24e),可以得到:
Figure PCTCN2019081475-appb-000056
由式(34),(46)及(47),可以得到平稳分布:
Figure PCTCN2019081475-appb-000057
综上,将所求的平稳分布式(48)代入式(35),就可以求得给定资源分配比及转发阈值条件下N G,γ的取值。综上,求解平稳分布的算法如表1所示。
表1求解平稳分布的算法
Figure PCTCN2019081475-appb-000058
5.2应用修正的差分演化算法求解最优解
式(24)所描述的优化问题是一个带约束的混合整数非线性规划问题,由于该问题同时含有连续型变量及离散型变量,求解较为困难,因此,我们采用修正的差分演化算法进行求解。差分演化算法是一种基于实数编码的,在连续空间进行全局优化的启发式算法,通过群体间个体的合作与竞争所产生的群体来智能的指导优化搜索的方向,具体包含种群初始化,变异,交叉,选择4个操作。接下来分别介绍差分演化中这4个操作的具体内容。
(1)种群初始化
假设种群大小为ps,将该种群的第l代的第m个个体的取值表示为:
Figure PCTCN2019081475-appb-000059
则种群初始化的具体操作为根据每个分量的取值范围随机均匀的产生第一代种群:
Figure PCTCN2019081475-appb-000060
其中,
Figure PCTCN2019081475-appb-000061
在开区间(0,1)上随机均匀生成,
Figure PCTCN2019081475-appb-000062
在1~γ U之间的整数中随机均匀生成。由此,我们便可以对当代种群的每个个体通过变异,交叉,选择等操作生成下一代群体,如此循环,不断迭代,使种群不断进化,从而找到最优解。
(2)变异
变异操作指的是基于当代种群中随机选择的个体的差值对每个父代个体生成相应的变异个体,假设当前种群为第l代种群,则对于第m(m=1,2,…,ps)个父代个体
Figure PCTCN2019081475-appb-000063
其通过变异操作所产生的变异个体为:
Figure PCTCN2019081475-appb-000064
其中,s 1,s 2,s 3——种群中随机选取的3个个体且s 1≠s 2≠s 3≠m,F——变异因子,一般F∈(0,2)。
(3)交叉
交叉操作指的是根据一定的概率,将父代个体
Figure PCTCN2019081475-appb-000065
与变异个体
Figure PCTCN2019081475-appb-000066
在每个分量上分别进行交叉从而生成临时个体。第l代种群的第m个个体通过该操作所产生的临时个体的每个分量为:
Figure PCTCN2019081475-appb-000067
其中,
Figure PCTCN2019081475-appb-000068
的第n个分量,r 1,n(n=1,2)——在区间[0,1]间随机均匀
Figure PCTCN2019081475-appb-000069
个分量。
选择操作指的是在父代个体
Figure PCTCN2019081475-appb-000070
与临时个体
Figure PCTCN2019081475-appb-000071
中选择最优的个体作为下一代个体。对于最大化问题,就是选择使目标函数更大的作为下一代,父代个体
Figure PCTCN2019081475-appb-000072
通过该操作所产生的下一代个体为:
Figure PCTCN2019081475-appb-000073
其中,g()表示优化问题中的目标函数。
以上这4步操作是最基本的差分演化算法的过程介绍,适用于求解无约束连续型变量的优化问题,由于我们的优化问题是带约束的混合整数非线性规划问题,所以需要对上述算法作以修正。具体包含3个内容,分别为约束函数的处理,整型变量γ的处理,边界违反情况的处理,接下来分别 介绍这3个内容。
(1)约束函数的处理
在修正的差分演化算法中,我们采用罚函数处理函数约束,将我们的优化问题转化为无约束优化问题。将约束函数式(24f)改写为:
Figure PCTCN2019081475-appb-000074
因为
Figure PCTCN2019081475-appb-000075
且π i=-a iπ γ-1>0(i=0,...γ-1)所以
Figure PCTCN2019081475-appb-000076
Figure PCTCN2019081475-appb-000077
Figure PCTCN2019081475-appb-000078
所以若π γ>0,就可以满足0<π i<1,(i=0,...γ)这个约束,因此罚函数中还需考虑-π γ<0这个约束,于是我们的罚函数可以表示为:
g(β,γ)=T(β,γ)-σ[max(0,f(β,γ))+max(0,-π γ)]    (55)
其中,σ——惩罚因子,取足够大的正数。当(β,γ)满足约束条件时,我们有g(β,γ)=T(β,γ);当(β,γ)不满足约束条件时,我们有g(β,γ)=T(β,γ)-σ[max(0,f(β,γ))+max(0,-π γ)]。上述罚函数表明(β,γ)离约束条件越远,惩罚越大,即罚函数在搜索到不可行点时,g(β,γ)会变得很小,目标函数值会变得很小。
基于以上分析,我们的带约束优化问题式(24)可以转化为下述无约束优化问题:
Figure PCTCN2019081475-appb-000079
s.t.0<β<1                   (56b)
γ∈N +,1≤γ≤γ U           (56c)
Π=ΠP            (56d)
Figure PCTCN2019081475-appb-000080
(2)整型变量γ的处理
在我们的修正差分演化算法中,我们采用四舍五入取整法处理整型变量γ,即在差分演化算法的每次迭代中,搜索时我们将γ看作实数,但是在求平稳分布,N G,γ及目标函数g(β,γ)之前,对γ四舍五入取整。
(3)边界违反情况的处理
我们可以看出差分演化算法的变异操作本身不具有封闭性,变异个体可能存在边界违反的情况,即某分量的取值可能超出边界,因此当某一分量取值超出边界时,我们在其取值范围内随机生成一个数来代替该分量。
基于以上讨论,我们的修正的差分演化算法如表2所示。按照该算法我们就可以求得最优资源 分配比β *及最优转发阈值γ *
表2修正的差分演化算法
Figure PCTCN2019081475-appb-000081
6.仿真结果
(1)仿真参数设置如表3所示:
表3仿真参数设置
参数
MTCD流量到达模型 均匀分布
MTCG总数 20
可用PA数 54
所有MTCD到达所需总RAO数 2000
总资源块数Q M 3600
链路1传输一个数据包所需资源块数δ 1 20
链路2转发一个数据包所需资源块数δ 2 5
网关缓存大小上界γ U 50
每个设备的最大重传次数 10
退避窗口所需总RAO数 4
种群大小ps 40
变异因子F 1
交叉概率cp 0.6
惩罚因子σ 10^6
差分演化算法迭代次数L 100
截取时间开始时隙 第750个RAO
截取时间结束时隙 第1500个RAO
在这一部分,我们通过仿真来评估我们所提方案的系统性能,参数的具体设计如表3所示。为了证明我们所提出的基于分组的随机接入与数据传输方案的性能优越性,我们用带有传统ACB的传统正交随机接入方案做以对比。
图3所示为2种方案下总服务时间随系统中每组MTCD数的变化曲线,这里总服务时间定义为系统服务完所有MTCD所需的总RAO数,所有设备被服务完指的是所有设备要么在最大传输次数内传输成功,要么由于超过最大重传次数而传输失败。可以看出2种方案下,总服务时间随每组MTCD数的增加而增加,但是本文分组方案的增长速度较慢,且本发明设计的方案相对于传统方案可以大大减少总服务时间。因为本发明考虑到MTCD数量巨大且单个数据包较小的特性,传统方案用于传输数据的资源较多,网关传输速率较高等原因,在传统方案中引入网关,设计了双跳mMTC网络架构,用马尔科夫链分析了网关缓存状态的平稳分布,并在此基础上对系统进行了优化,以最大化系统吞吐量,从而提升了系统性能,大大减少了总服务时间。
图4所示为2种方案下总吞吐量随系统每组MTCD数的变化曲线,这里总吞吐量定义为在最大传输次数内成功传输的MTCD总数,该图中分别画出了总服务时间内的总吞吐量及截取时间段内的总吞吐量。可以看出2种方案下,总服务时间内的总吞吐量相当,也就是说2种方案下,几乎所有设备都可以在最大重传次数内传输成功,但是由图3可以看出传统方案需要花费更多时间。另外,截取时间段内的总吞吐量在2种方案下有较大差异,随着每组MTCD数的增加,本发明的方案可以在截取时间段内服务更多设备,具体原因同图3。
图5所示为2种方案下平均吞吐量随系统每组MTCD数的变化曲线,这里平均吞吐量定义为平均一个RAO所能服务的MTCD数,即在最大传输次数内成功接入的MTCD数与服务所需总RAO数的比值,该图中分别画出了总服务时间内的平均吞吐量及截取时间段内的平均吞吐量。可以看出2种方案下,总服务时间内及截取时间段内的平均吞吐量相当,但是本发明的方案较传统方案可以大大提高平均吞吐量。具体原因与图3类似。
图6所示为2种方案下成功接入概率随系统每组MTCD数的变化曲线,该图中分别画出了总服务时间内及截取时间段内的成功接入概率,这里成功接入概率定义为所定义时间段内接入成功的MTCD数与传输的总MTCD数的比值。可以看出2种方案下,总服务时间段内接入成功概率相当,但是截取时间段内本发明的方案成功接入概率更高。具体原因与图3类似。
图7所示为2种方案下成功设备的平均接入时延随系统每组MTCD数的变化曲线,这里平均接入时延定义为平均一个成功MTCD从初次激活到基站接收成功所需的RAO数。可以看出2种方案下,在总服务时间内,本发明的方案可以大大减少平均时延,在截取时间段内,本发明的方案同样也减少了成功设备的平均接入时延。具体原因与图3类似。
因此综上可知,本发明提出的基于分组的随机接入与数据传输方案更适用于mMTC网络,对于整个系统,可以大大提升吞吐量,减小系统总服务时间,对于每个MTCD,可以减小平均接入时延,增加成功接入概率。
以上内容是对本发明进行的详细说明,不能认定本发明的仅限于此,对于本发明所属技术领域的普通技术人员来说,在不脱离本发明构思的前提下,还可以做出若干简单的推演或替换,都应当视为属于本发明由所提交的权利要求书确定专利保护范围。

Claims (5)

  1. 大规模MTC网络中一种基于分组的随机接入与数据传输方法,其特征在于,包括以下步骤:
    (1)在mMTC网络中,引入MTC网关MTCG,让每个网关管理若干个设备,分别对应于一个MTCD组,将用于数据传输的资源按照修正差分演化算法所求的最优资源分配比β *分为2部分:即链路1与链路2,将链路1中的资源均分给每组,每组的资源按照MTCD传输一个数据包所需的资源块数划分为U 1个信道,根据修正差分演化算法所求的最优转发阈值γ *及网关转发一个数据包所需的资源块数将链路2划分为U 2个信道;
    (2)MTC网关向组内激活设备广播类ACB因子,激活设备进行类ACB检测,通过类ACB检测的组内设备在U 1个正交信道中随机选择一个信道将数据包传输给相应的MTC网关;
    (3)MTCG检测U 1个正交信道,其中只有某个信道被一个设备选择时,MTCG才能正确检测出对应的数据包,对于检测失败的数据包其对应设备进行均匀退避;
    (4)网关根据缓存余量接收成功检测的数据包,若缓存余量大于等于检测成功的数据包数,则所有检测成功的数据包被网关成功接收并存储在缓存队列中,否则网关根据缓存余量大小在检测成功的数据包中随机选择若干个并存储在缓存队列中,使得缓存队列长度达到最优转发阈值γ *
    (5)缓存队列长度达到最优转发阈值γ *的MTCG向BS发起接入请求,通过基于竞争的随机接入四消息握手过程接入基站,接入成功的MTCG进行下一步,接入失败的MTCG在下一个随机接入机会再次发起接入请求;
    (6)基站根据链路2的正交信道数随机选择接入成功的MTCG并分配信道供网关转发缓存中的数据包,若U 2大于等于接入成功的MTCG数,则所有接入成功的MTCG都可以成功转发数据包,否则基站在这些MTCG中随机选择U 2个并分配信道供相应网关在链路2中转发数据包,未获得链路2信道的MTCG在下一个随机接入机会再次发起接入请求。
  2. 根据权利要求1所述的大规模MTC网络中一种基于分组的随机接入与数据传输方法,其特征在于,所述步骤(1)中最优资源分配比β *和最优转发阈值γ *,首先通过马尔可夫理论分析出网关缓存队列长度的状态转移概率矩阵,对网关缓存占有量的变化过程进行数学建模,得到每一个随机接入机会中发起接入请求的网关数,即
    Figure PCTCN2019081475-appb-100001
    π i,(i=0,1,...γ)表示系统达到平稳状态时网关缓存大小在每一RAO结束时刻为i的概率,N G表示网关数,P G,t(i=0,1,2,...,U 1)表示一个MTCG在一个RAO成功检测i(i=0,1,2,...,U 1)个包的概率,从而得到每一个随机接入机会的吞吐量,即T(β,γ)=N G,γ·P s·γ,P s表示网关成功转发的概率,结合资源分配比网关缓存上界以及平稳分布的条件构造吞吐量的优化问题,利用修正差分演化算法求得最优转发阈值γ *和最优资源分配比β *
  3. 根据权利要求1或2所述的大规模MTC网络中一种基于分组的随机接入与数据传输方法,其特征在于,所述的吞吐量的优化问题如下:
    Figure PCTCN2019081475-appb-100002
    Q M表示系统在一个随机接入机会中用于传输数据的总资源块数,δ 2表示网关转发一个数据包所需的资源块数,γ U表示网关实际缓存上界,Π表示系统的平稳分布,P表示网关缓存占有量的转移概率矩阵,N PA表示前导码数。
  4. 根据权利要求2所述的大规模MTC网络中一种基于分组的随机接入与数据传输方法,其特征在于,所述的修正差分演化算法是1)利用罚函数将原优化问题转化为无约束优化问
    Figure PCTCN2019081475-appb-100003
    其中,g(β,γ)表示罚函数;2)在算法迭代过程中,将整型变量转发阈值γ看作实型变量再求目标函数之前将实型变量γ四舍五入取整,3)由于差分演化算法变异操作不封闭,当通过变异操作得到的资源分配比β和转发阈值γ超出取值范围时,在取值范围内随机生成以取代超出边界的变量。
  5. 根据权利要求4所述的大规模MTC网络中一种基于分组的随机接入与数据传输方法,其特征在于,所述罚函数g(β,γ)=T(β,γ)-σ[max(0,f(β,γ))+max(0,-π γ)],其中σ为惩罚因子。
PCT/CN2019/081475 2019-03-14 2019-04-04 大规模mtc网络中一种基于分组的随机接入与数据传输方法 Ceased WO2020181597A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910194046.4 2019-03-14
CN201910194046.4A CN109803246B (zh) 2019-03-14 2019-03-14 大规模mtc网络中一种基于分组的随机接入与数据传输方法

Publications (1)

Publication Number Publication Date
WO2020181597A1 true WO2020181597A1 (zh) 2020-09-17

Family

ID=66563566

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/081475 Ceased WO2020181597A1 (zh) 2019-03-14 2019-04-04 大规模mtc网络中一种基于分组的随机接入与数据传输方法

Country Status (2)

Country Link
CN (1) CN109803246B (zh)
WO (1) WO2020181597A1 (zh)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110602788B (zh) * 2019-08-27 2023-05-02 国网重庆市电力公司电力科学研究院 一种窄带物联网上行时频资源优化方法及可读存储介质
CN110636476B (zh) * 2019-09-23 2022-09-20 南京邮电大学 用于缓解物联网大规模m2m通信中拥塞问题的方法
CN111294775B (zh) * 2020-02-10 2021-04-20 西安交通大学 一种大规模mtc与h2h共存场景中基于h2h动态特性的资源分配方法
CN112004213B (zh) * 2020-08-28 2022-07-29 南京邮电大学 一种mMTC分层系统中基于干扰图的资源分配方法
CN112996081A (zh) * 2021-02-04 2021-06-18 慕熠物联网技术(湖州)有限公司 一种基于时间敏感算法的窄带无线通信系统终端节点
CN117692961B (zh) * 2024-02-04 2024-04-26 北京邮电大学 低轨卫星物联网随机接入拥塞控制方法和装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106851732A (zh) * 2010-08-12 2017-06-13 英特尔公司 机器类型通信数据的数据处理方法、装置及系统
CN108282821A (zh) * 2018-01-23 2018-07-13 重庆大学 一种物联网通信中面向巨连接的基于分组的拥塞控制接入方法
WO2018234851A1 (en) * 2017-06-21 2018-12-27 Telefonaktiebolaget Lm Ericsson (Publ) Random access procedure for massive mtc devices
CN109218004A (zh) * 2018-10-26 2019-01-15 南京邮电大学 基于SCMA mMTC系统分层机器类通信的子载波及功率分配方法

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
HK1251765A1 (zh) * 2015-06-25 2019-02-01 苹果公司 5g网络中的动态分区资源分配
CN108848482B (zh) * 2018-06-12 2021-03-16 南京邮电大学 基于mMTC分层接入框架的资源分配方法
CN108882301B (zh) * 2018-07-25 2020-07-28 西安交通大学 大规模m2m网络中基于最优功率退避的非正交随机接入方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106851732A (zh) * 2010-08-12 2017-06-13 英特尔公司 机器类型通信数据的数据处理方法、装置及系统
WO2018234851A1 (en) * 2017-06-21 2018-12-27 Telefonaktiebolaget Lm Ericsson (Publ) Random access procedure for massive mtc devices
CN108282821A (zh) * 2018-01-23 2018-07-13 重庆大学 一种物联网通信中面向巨连接的基于分组的拥塞控制接入方法
CN109218004A (zh) * 2018-10-26 2019-01-15 南京邮电大学 基于SCMA mMTC系统分层机器类通信的子载波及功率分配方法

Also Published As

Publication number Publication date
CN109803246B (zh) 2020-08-18
CN109803246A (zh) 2019-05-24

Similar Documents

Publication Publication Date Title
CN109803246B (zh) 大规模mtc网络中一种基于分组的随机接入与数据传输方法
Osti et al. Analysis of PDCCH performance for M2M traffic in LTE
CN109905918B (zh) 一种基于能效的noma蜂窝车联网动态资源调度方法
CN108882301B (zh) 大规模m2m网络中基于最优功率退避的非正交随机接入方法
Wang et al. When to preempt? Age of information minimization under link capacity constraint
Althumali et al. Dynamic backoff collision resolution for massive M2M random access in cellular IoT networks
CN109890085B (zh) 一种分优先级机器类通信随机接入退避参数确定方法
Pacheco-Paramo et al. Delay-aware dynamic access control for mMTC in wireless networks using deep reinforcement learning
Ruan et al. Deep neural network supervised bandwidth allocation decisions for low-latency heterogeneous e-health networks
Gedikli et al. Deep reinforcement learning based flexible preamble allocation for RAN slicing in 5G networks
WO2021057799A1 (zh) 一种基于联合时频分集的免授权上行传输方法
CN109963272B (zh) 一种面向差异化mtc网络中的随机接入方法
Orim et al. Random access scheme for machine type communication networks using reinforcement learning approach
CN110113720B (zh) 基于acb机制的组寻呼拥塞控制方法
Kim et al. Delay analysis and optimality of the renewal access protocol
CN111970703A (zh) NB-IoT中的一种上行通信资源优化的方法
Bui et al. Free access distributed queue protocol for massive cellular-based M2M communications with bursty traffic
Chan et al. On the stability and optimal decentralized throughput of CSMA with multipacket reception capability
Bouzouita et al. A random access model for M2M communications in LTE-advanced mobile networks
Orim et al. Random access sliced 5G-Advanced network for machine type communication
Alavikia et al. Random multiple data packets transmission scheme in LTE-based machine-type communications
Cárdenas et al. Collision-Aware Random Access Procedure for 5G/6G Mobile Networks in mIoT Scenarios
Jian et al. An traffic intensity based media access control protocol with priorities
Yi et al. Intelligent Multi-link EDCA Optimization for Delay-Bounded QoS in Wi-Fi 7
KR102758317B1 (ko) 5g 에지 네트워크의 지능형 미디어 포렌식 및 트래픽 처리 방법

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19919231

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19919231

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 19919231

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 25/05/2022)

122 Ep: pct application non-entry in european phase

Ref document number: 19919231

Country of ref document: EP

Kind code of ref document: A1