Cellular5GAccessforMassiveIOT.rar_5G_5GCommunications

共1个文件

pdf：1个

版权申诉

massive

95 浏览量 2022-07-15 20:04:18 上传评论收藏 1.98MB RAR 举报

资源推荐

资源详情

资源评论

收起资源包目录

Cellular 5G Access for Massive IOT.rar （1个子文件）

Cellular 5G Access for Massive IOT.pdf 2.41MB

17 Cellular 5G Access for Massive

Internet of Things

Germán Corrales Madueño, Nuno Pratas,

Cedomir Stefanovi

c, and Petar

Popovski

17.1 Introduction to the Internet of Things (IoT)

The IoT refers to the paradigm of physical and virtual “things” that communicate

and collaborate over the Internet, with or without human intervention. The spectra of

things that may be connected within the IoT ranges from complex machines, such as

aircraft and cars, to everyday appliances, such as consumer refrigerators, and very

simple devices such as humidity sensors. The emphasis of the IoT is on services,

which represent the primary driver for interconnecting things. Examples of IoT services

include micro-climate monitoring of homes, asset tracking during transportation, and,

on a larger scale, controlling the power consumption of all the refrigerators in a country

depending on the load. Current and forecast market evaluations (such as Cisco’s forecast

of a $14.4 trillion global IoT market by 2022 [1]) show that the IoT has a huge revenue

potential, to be shared between operators, service providers, hardware vendors, and

testing-solutions vendors. Thus, it is not surprising that the IoT is currently one of

the hottest topics in the telecommunications world, endorsed by both industry and

academia.

A term closely related, but not identical to IoT is machine-to-machine (M2M)

communications, or, in the Third Generation Partnership Project (3GPP) terminology,

machine-type-communications (MTC). M2M communications refer to the concept in

which machines (i.e., standalone devices) communicate with a remote server without

human intervention. “M2M can be considered as the plumbing of IoT” [2] or, more

formally stated, M2M communications are the key enabler of IoT services. A natural

question that arises is how well the existing networking solutions and technologies can

serve as the basis for M2M communications and, more broadly, IoT services and,

when they cannot support them, how to design other, suitable connectivity solutions.

These questions have in recent years instigated a signiﬁcant body of research and

development by industry, standardization bodies, and academia. The general conclusion

is that the existing technologies, in their present form, cannot efﬁciently support M2M

communications. The reason is that existing communication systems, particularly in

the wireless domain, are designed to efﬁciently support human-type communications

(HTC), such as web browsing, voice calls, and video streaming, where high data rates

are essential but the volume of users that simultaneously require service is far beyond

available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316771655.018

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 15 Aug 2017 at 14:54:48, subject to the Cambridge Core terms of use,

Cellular 5G Access for Massive Internet of Things 381

the expected number of interconnected devices. In contrast, M2M communications are

characterized by the transmission and reception of small data packets from a potentially

massive number of devices. This fundamentally changes the goal of the design

of the communication system from increasing link capacity to the implementation

of low-overhead signaling protocols and handling a large number of simultaneous

connections. In summary, the identiﬁed deﬁciencies of the existing communication

technologies have led to development of new, typically proprietary solutions [3, 4]

or to enhancing the current technologies [5], and these are competing to become the

communication protocol of choice for M2M communications and, therefore, the IoT.

Cellular technologies are seen as viable candidates in this race, owing to their

maturity and worldwide availability and the use of reserved spectrum. Nevertheless,

the main obstacle to the efﬁcient support of M2M communications in cellular networks

is that such networks are not designed to support a large number of simultaneous

connection attempts. Speciﬁcally, owing to the sporadic nature of data transmission

in M2M communications, where a device only occasionally transmits a short report,

very often a device has to ﬁrst establish a connection with an access point, i.e.,

base station, using a connection establishment procedure. This procedure involves

a considerable amount of signaling exchange and starts in a slotted ALOHA-based

manner, involving several bottlenecks when a massive number of devices are active

simultaneously [6]. In such cases, the connection establishment will fail for most of

the devices, resulting in outages, i.e., service deprivation, which makes the cellular

access mechanisms for M2M/IoT communications extremely important. Furthermore,

many IoT devices will be placed inside buildings and basements, involving signiﬁcant

a decrease in signal power due to penetration losses. This calls for a technology

that supports extended coverage. Some IoT devices are expected to operate for more

than 10 years without charging or replacing the battery, thereby requiring ultra-low

power consumption. Finally, in addition to supporting massive IoT, the next generation

of cellular technologies should also support another class of IoT devices that offer

mission-critical communication. These are characterized by requirements for ultra-high

reliability and low latency. Although massive M2M is usually treated separately

from mission-critical M2M communication, in this chapter we will focus mainly on

massive M2M, occasionally referring to the technology issues for mission-critical M2M

communications.

17.2 IoT Trafﬁc Patterns in Network Access

The IoT is characterized by a wide application range, for example monitoring and

control of household appliances, home security, elderly care, e-health, smart metering,

monitoring and control of smart grids, industrial processes, and trafﬁc and automotive

applications. All of these applications share common features that impact access to the

cellular network [7]:

• Small data transmissions, with a packet/report size below 1000 bytes.

available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316771655.018

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 15 Aug 2017 at 14:54:48, subject to the Cambridge Core terms of use,

382 Germán Corrales Madueño et al.

• Uplink-dominated transmissions.

• Typically, delay-tolerant transmissions for a regular/periodic mode of reporting.

• Service priority and low delays for an alarm mode of reporting.

• Low-mobility devices, which are in most cases sensors or controllers deployed at a

speciﬁc position.

• Low-complexity and low-power transceivers with limited processing capabilities and

lack of software updates, in order to achieve extremely low cost.

• Strict requirements on energy consumption, as it is assumed that the device battery is

not rechargeable, while the operational time should be more than 10 years.

• A potentially very large number of devices in a cell.

Moreover, there are two dominant trafﬁc patterns that can be observed in all

potential IoT applications: asynchronous and synchronous reporting. In asynchronous

reporting, the report arrivals are not correlated across different devices, i.e., the

devices report in an independent and random manner. For example, asynchronous

reporting arises in the case where earthquake sensors periodically inform a remote

server that they are operational. Although the cause of the reporting is the same,

the likelihood of multiple devices reporting at the same time is almost negligible.

On the other hand, in synchronous reporting, there is a strong correlation in the

trafﬁc patterns across devices such that a considerable number of devices become

active simultaneously or nearly simultaneously, typically owing to an external event

that triggers these devices. For example, synchronous reporting arises in the case

where earthquake sensors are triggered by an actual earthquake, when a multitude

of devices will try report the detected seismic activity to the remote server almost

simultaneously.

Obviously, the asynchronous/synchronous nature of trafﬁc patterns is strongly related

to the reporting mode of the IoT devices. Speciﬁcally, there are three main reporting

modes, depicted in Figure 17.1: (i) periodic reporting, (ii) on-demand reporting, and

(iii) alarm reporting, where the ﬁrst two modes give rise to the trafﬁc patterns of

asynchronous reporting and the last mode to synchronous reporting.

Periodic reporting occurs when reports are sent to or exchanged with the remote

server under standard operation conditions in some monitoring application. Despite the

Time

Time Time

(a) (b)

(c)

Traffic

Figure 17.1 Typical trafﬁc patterns in the IoT: (a) periodic reporting, (b) on-demand reporting,

and (c) alarm reporting.

available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316771655.018

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 15 Aug 2017 at 14:54:48, subject to the Cambridge Core terms of use,

Cellular 5G Access for Massive Internet of Things 383

Table 17.1 Possible cellular scenario with IoT applications [8, 10]. Overhead

due to IP packets not included.

IoT application RI min Payload (bytes) Number of devices

Smart meters 1, 5, 15, > 60 100–1000 20 911

Home security systems 10 20 4 647

Elderly sensor devices 1 128 465

Credit machines in groceries 2 24 108

Credit machines in shops 30 24 1 650

Roadway signs 0.5 1 4 444

Trafﬁc lights 1 1 540

Trafﬁc sensors 1 1 540

name “periodic,”

∗

periodic reporting is commonly modeled as a uniform or Poisson

process [8] in terms of the distribution of the times of report arrivals over the reporting

devices. In the latter case, it assumed that the intensity of the process is λ

= 1/RI,

where RI is the reporting interval of the device in seconds [9]. The expected number of

report arrivals 

(τ ) for a single device during an interval of τ is

E[

(τ )]=λ

τ , (17.1)

while the probability of k report arrivals during τ is

P[

(τ ) =k]=

(λ

τ)

−λ

. (17.2)

The aggregate of the trafﬁc arrivals over devices is also Poisson distributed, with an

(aggregate) intensity of Nλ

, where N is the number of reporting devices in a cell; it

is assumed that the trafﬁc arrivals are independent and identically distributed (i.i.d.)

processes over the reporting devices involved in the same application. The expected

number of aggregate arrivals during τ is simply Nλ

τ . The number of reporting devices

in an IoT application per cell varies from source to source, spanning a range from several

thousand [10] to up to 300 000 per cell in some use cases foreseen for ﬁfth generation

(5G) systems in coming years [11]. This number also depends on the type of cell being

considered, i.e., suburban or urban. Table 17.1 shows a potential scenario in terms of

the expected number of devices (both commercial and in-home) and their payloads

in a sub-urban cell of radius 1500 m with three sectors [8, 10]. The main parameters

considered are the average message size in bytes, the average message transaction rate

per second (i.e., the intensity), and the trafﬁc distribution. Smart meters represent a

special case, with reporting intervals ranging from 1 minute to several hours.

On-demand reporting occurs when the end user or the application server requests

information from the device, for example, an energy consumption report demanded by

∗

In fact, it would be more suitable to use the term “regular reporting,” since it is implicitly assumed that the

devices involved in the same application are conﬁgured or initialized such that their reporting periods are

somewhat randomized in time. If this is not the case, then multiple devices may end up using the same time

instants for reporting, thereby making the reporting pattern synchronous and thus increase the chances of

congesting the network.

available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316771655.018

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 15 Aug 2017 at 14:54:48, subject to the Cambridge Core terms of use,

384 Germán Corrales Madueño et al.

a customer in order to learn about the amount of energy consumed. The distribution

of trafﬁc arrivals for on-demand reporting can also be modeled as a Poisson process

[6], with an intensity λ

that is much lower than λ

. Thus, the total arrival intensity of

asynchronous trafﬁc per device is given by λ

=λ

+λ

, while the total arrival intensity

aggregated over the reporting devices is simply Nλ

. Finally, it should be noted that this

type of reporting is generally considered as delay tolerant [6, 12], as it can be delayed

for a certain amount of time without causing service disruption. Taking a smart metering

example, if an energy consumption report cannot be delivered at the time of generation,

it can be delayed until the next scheduled reporting interval.

Finally, alarm reporting corresponds to messages generated owing to observation of

an alarm condition. A common example of alarm reporting occurs when there is a power

outage event in the electrical grid, where potentially thousands of monitoring devices

will be affected and will try to report the outage simultaneously or near simultaneously.

The delay constraints of alarm reporting are more strict than for periodic or on-demand

reporting where the exact value of the tolerable delay depends on the particular

service/application. In another example, in smart power metering the tolerable delay

for the reporting of an outage is below 0.5 s, owing to the depletion of the integrated

battery (also known as last-gasp reporting) [10]. In [13], 3GPP proposed a model for

highly correlated trafﬁc arrivals, i.e., synchronous trafﬁc, where the arrival time interval

follows a Beta distribution,

p(t) =

α−1

(T −t)

β−1

α+β−1

B(α,β)

for 0 ≤ t ≤T, (17.3)

where α>0 and β>0 are shape parameters, B(α,β) is the Beta function [13], and T

is the activation period, i.e., the time period from the activation of the ﬁrst station until

the last device is activated. The values of the parameters α, β, and T suggested in [13]

are α = 3, β =4, and T = 10 s. The number of devices triggered during an interval τ is

given by



p(t) dt, (17.4)

where N is the total number of devices.

The assumption that alarm trafﬁc can be modeled via a Beta distribution seems to be

commonly accepted; however [13] does not provide any speciﬁc reasoning for choosing

this particular distribution or for choosing the particular values of the parameters α, β,

and T. Some insights are provided in [6], where a simulation of the physical propagation

of an alarm event, speciﬁcally, an earthquake, was carried out, and the results obtained

were compared with the results of the model proposed in [14]. The primary wave of an

earthquake propagates at different speeds depending on a number of factors, such as the

type of ground; in the simulated model it was assumed that the speed was v = 4km/s

[13]. The simulation assumed a circular cell with a radius r = 1500 m and that the

number of monitoring devices was N = 10000. The epicenter of the earthquake could

be located anywhere in the cell. Furthermore, it was expected that the likelihood of a

device being triggered by an event would decrease with distance. This was modeled via

available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316771655.018

Downloaded from https://www.cambridge.org/core. Columbia University Libraries, on 15 Aug 2017 at 14:54:48, subject to the Cambridge Core terms of use,

评论收藏

内容反馈

版权申诉