Signalling Storms in 3G Mobile Networks--使用HMM模型，参数： the key parameters of mobile user device behaviour that can lead to signalling storms，功耗角度（analysing signalling behaviour from an energy

consumption perspective,）见论文图，就是建立的HMM模型。

Abstract—We review the characteristics of signalling storms
that have been caused by certain common apps and recently
observed in cellular networks, leading to system outages. We
then develop a mathematical model of a mobile user’s signalling
behaviour which focuses on the potential of causing such storms,
and represent it by a large Markov chain. The analysis of this
model allows us to determine the key parameters of mobile user
device behaviour that can lead to signalling storms. We then
identify the parameter values that will lead to worst case load
for the network itself in the presence of such storms. This leads to
explicit results regarding the manner in which individual mobile
behaviour can cause overload conditions on the network and its
signalling servers, and provides insight into how this may be
avoided.

Impact of Signaling Storms on Energy Consumption and Latency of LTE User
Equipment——主要是探讨影响因素。

Abstract—Signaling storms in mobile networks, which congest
the control plane, are becoming more frequent and severe
because misbehaving applications can nowadays spread more
rapidly due to the popularity of application marketplaces
for smartphones. While previous work on signaling storms
consider the processing overhead in the network and energy
consumption of the misbehaving User Equipment (UE) only,
this paper aims to investigate how signaling storms affect both
the energy consumption and bandwidth allocation of normal
and misbehaving LTE UEs by constructing a mathematical
model which captures the interaction between the UE traffic
and the Radio Resource Control state machine and bandwidth
allocation mechanism at the eNodeB. Our results show that
even if only a small proportion of the UE population is
misbehaving, the energy consumption of the radio subsystem of
the normal UEs can increase significantly while the time spent
actively communicating increases drastically for a normal data
session. Moreover, we show that misbehaving UEs have to spend
an increasing amount of energy to attack the network when
the severity of the signaling storms increases since they also
suffer from the attacks.

previous
work which either investigate the impact of different RRCbased
attacks on the mobile network in terms of signaling
overhead and delay only [7]–[11] or the impact of traffic
behavior on the energy consumption of the misbehaving UEs
only without considering other UEs [12]–[15].

相关工作：

Several existing research papers [12]–[14] have investigated
the energy consumption of UEs due to different
application traffic patterns which can also lead to signaling
storms in LTE networks. In [12], the authors model the
DRX mechanism using a semi-Markov chain to obtain the
trade-off between power consumption and various DRX
parameters such as timeout, for bursty packet data traffic.
The analytical results were also verified against simulations.
[13] also investigates the same impact factors of

the DRX mechanism but it also takes into account the
various signaling messages that are exchanged during RRC
mode transitions. The most important contribution of [14]
is the measurement of the power consumption of LTE UEs
in different operational networks around the world during
different RRC and DRX states, which we use in this work.
In addition, the authors of [14] infer the different DRX
parameters used by operators from their power and traffic
measurements which they then use to build a power model
for a LTE UE so that they can compare the power and delay
performance of a LTE UE against a 3G and WIMAX UE.
Our previous work on signaling storms in the context of
the NEMESYS project [2], [17] has involved the mathematical
modeling, simulation and analysis of the impact of different
RRC-based signaling storms in 3G/UMTS networks
[9]–[11]. In our recent work, we also investigated methods
for the detection and mitigation of signaling storms through
the use of RRC timer’s adjustment and counters [16].

LTE里信令增加的原因——群发消息、永久在线！

2.1基于TAL 的寻呼策略导致信令量成倍上升
与3G 网络基于RU+AN 的寻呼机制不同，LTE 系统
的寻呼机制是基于TA list（跟踪区列表）实现的。所谓TA
list 是指将原来CDMA 系统LAC（location area code，位置
区码） 包括的基站归类为LTE 系统最小注册和寻呼单元，
MME 在下发寻呼消息时， 对终端最后注册的LAC 下的所

2.2 永久在线的功能进一步推高了信令流量
在系统设计上，3G 时代由于语音与数据是分开在两
张网络中承载的， 故并不需要电路域中的永久在线功能；
但到了4G 时代，没有了电路域，语音也在分组域上承载，
此时永久在线的功能就显得尤为重要， 因此在LTE 系统
中，终端一旦开机就进行附着和默认承载的建立（IP 地址
的获取），并且IP 地址的存活时间一直延续到终端去附着
或者是超出了PGW 设定的最大时间。众所周知，IP 地址的
存在是寻呼信令产生的前提，IP 地址存活时间越长， 网络
寻呼终端的信令量将越大，发生信令风暴的可能也就越大。

2.3 扁平化的系统架构加剧了信令风暴的危害——？？？去中心吗？？？
与3G 系统架构不同的是，LTE 的系统架构少了中央
控制器节点RNC，无线基站直接对接核心网元MME。由于
MME 并不知道终端的邻区信息，因此，3G 网络中基于RU
和neighbor list 的寻呼在LTE 系统中不可行， 阻碍了寻呼
范围的缩小，使得寻呼信令无法降低。
同时，由于MME 下挂的小区数远高于（RNC），以广东
电信LTE 网络为例， 全省共建设两套MME， 组成MME
pool，管理上万个eNB，因此，一旦发生信令风暴问题，MME
处理能力将下降甚至瘫痪，影响面更广，破坏性更大。

2.4 带宽拓展推动移动互联网应用的爆发式增长
根据爱立信消费者实验室的研究报告， 如图3 所示，
到2018 年， 全球LTE 网络的用户数量将超过20 亿户，年
复合增长率75%。与此同时，QQ、微信等OTT 业务的广泛
使用，使得点对点寻呼、点对多点的组呼更为频繁，网络承
载的信令将出现几何级增长。

from:《LTE 网络信令风暴风险分析与对策研究》