A First Look at Unstable Mobility Management in Cellular Networks - - PowerPoint PPT Presentation

▶
a first look at unstable mobility management in cellular
SMART_READER_LITE
LIVE PREVIEW

A First Look at Unstable Mobility Management in Cellular Networks - - PowerPoint PPT Presentation

Nov 10, 2022 355 likes •648 views

A First Look at Unstable Mobility Management in Cellular Networks Yuanjie Li 1 Jiaqi Xu 2 Chunyi Peng 2 Songwu Lu 1 1 University of California, Los Angeles 2 The Ohio State University HotMobile16 Ubiquitous Cellular Network Access Cellular

slide-1
SLIDE 1

A First Look at Unstable Mobility Management in Cellular Networks

Yuanjie Li1 Jiaqi Xu2 Chunyi Peng2 Songwu Lu1

1University of California, Los Angeles 2The Ohio State University

HotMobile’16

slide-2
SLIDE 2

Ubiquitous Cellular Network Access

Chunyi Peng HotMobile'16 2

7.9+ billion

in 2015

Cellular Networks

slide-3
SLIDE 3

Mobility Management (MM) via Handoff

Chunyi Peng HotMobile'16 3

Cellular Networks § Seamless connectivity (via switching the serving cell)

¨ Each cell: limited radio coverage

slide-4
SLIDE 4

Desirable Handoff: Stability

Chunyi Peng HotMobile'16 4

Cell 1 Cell 3 Cell 2

1 2 3 4

Handoff movement

§ Converge to certain cell given an invariant setting Why desirable? § Handoff comes at a cost

¨ Multi-round signaling

exchange

¨ Service

disruption/degradation

slide-5
SLIDE 5

Desirable Handoff

Chunyi Peng HotMobile'16 5

Cell 1 Cell 3 Cell 2

1 2 3 4

Handoff movement

Cell 1 Cell 3 Cell 2

1 2 3 4 5 6

Problematic Handoff

§ Instability (persistent loop):

¨ C1->C2->C3->C1->C2->C3…

§ Stability

¨ Converge to certain cell

slide-6
SLIDE 6

This Work: Instability in Mobility Management

  • Q1: Does it exist in real networks?
  • Q2: Why unstable?
  • Q3: How to identify such risk?

Chunyi Peng HotMobile'16

Caused by fundamental (persistent) conflicts in policy not by transient factors (radio dynamics etc)

slide-7
SLIDE 7

Q1: Does unstable MM exist in reality?

  • Unfortunately, Yes!

Chunyi Peng HotMobile'16 7

slide-8
SLIDE 8

10 11 12 13 14 15

3 2 1 20 40 60 Serving Cell (minute)

3-Cell Loop Example

§ Static, 40hr-loop

Chunyi Peng HotMobile'16 8

Cell 1 Cell 3 Cell 2

1hr trace

Loop every few mins (90% loops in 200s)

Cell1: 4G Cell2: Femtocell (3G) Cell3: 3G

slide-9
SLIDE 9

Negative Impacts in Real-world

§ Excessive signaling

  • verhead (2-8x)

Chunyi Peng HotMobile'16 9

2K 4K 6K 8K 1 2 3loop

  • num. msgs per hour

§ Performance degradation (10+ fold slowdown)

30 60 90 120 150 180 1 2 3loop Downloading time (s)

20 40 60 80 1 2 3loop Web loadtime (s) 5s

8.5x 3.5x 2.2x cnn 5MB file 12s 3s 76s 180s

§ Hurt both carriers and users

slide-10
SLIDE 10

Q2: Why is MM unstable?

Chunyi Peng HotMobile'16 10

slide-11
SLIDE 11

Distributed Nature of Handoff

§ Each handoff: trigger-decision-execution phases § Sequence of handoffs

Chunyi Peng HotMobile'16 11

Cell1

2

decision rules paras meas Cell 2

3 1

Handoff decision@C1

… …

5

decision rules’ paras’ meas’

4

Handoff decision@C2

slide-12
SLIDE 12

Handoff for Versatile Demands

Chunyi Peng HotMobile'16 12

C1

2

decision rules paras meas

Handoff decision@C1

§ Seamless connectivity § Voice/data support § Performance improvement § Load balancing § …

Each individual handoff: OK ≠ The interplay among multiples: OK

slide-13
SLIDE 13

3-Cell Loop Example

Chunyi Peng HotMobile'16 13

Cell 1 Cell 3 Cell 2

Cell1: 4G Cell2: Femtocell (3G) Cell3: 3G

Rule/preference configuration@C1 (4G)

§ C2 (Femto) > C1 (4G) for offloading § C1 (4G) > C3 (3G) for higher-speed

C1: C1à C2 @C2 (3G Femto) § Best radio strength with same preferences for all cells C2: C2à C3 @C3 (3G)

§ C1 (4G) > C3 (3G) for higher-speed

C3: C3à C1

slide-14
SLIDE 14

From Example to Generalization

§ Each handoff decision: t = Fs(s, C)

¨ s: serving cell ¨ C: set of candidate cells ¨ Fs: decision function for serving cell s ¨ t: target cell

§ The sequence of handoff decisions

s → Fs(s) → · · · ci → [ci+1 = Fci (ci)] → · · · , ci ∈ C.

Chunyi Peng HotMobile'16 14

slide-15
SLIDE 15

From Example to Generalization

§ Instability = No convergence

¨ e.g., persistent loop: c → · · · ci →ci+1 → · · · c.

§ [Necessary stability condition] there exists at least one t, s.t. ∃t ∈ C, t = Ft(t, C) § [Necessary and sufficient condition] (1) ∃t ∈ C, t = Ft(t, C); (2) there exists a handoff path from the initial cell s to the desirable t

Chunyi Peng HotMobile'16 15

slide-16
SLIDE 16

Q3: How to detect possible instability?

Chunyi Peng HotMobile'16 16

slide-17
SLIDE 17

MMDIAG

§ In-device diagnosis

¨ Carriers: reluctant to provide network-side MM info

§ Two-phase: analyzer and validation

Chunyi Peng HotMobile'16 17

MM Automata

Config. Collector Scenario Emulator static dynamic Instability Analyzer Counter examples Empirical Validation

slide-18
SLIDE 18

MMDIAG

Chunyi Peng HotMobile'16 18

Config. Collector Scenario Emulator static dynamic Instability Analyzer Counter examples Empirical Validation

§ Model based on 3GPP spec § Decision logic, configuration parameters and runtime

  • bservation (scenario)

MM Automata

slide-19
SLIDE 19

MMDIAG

Chunyi Peng HotMobile'16 19

MM Automata

Config. Collector Scenario Emulator static dynamic Instability Analyzer Counter examples Empirical Validation

§ Model based on 3GPP spec § Decision logic, configuration parameters and runtime

  • bservation (scenario)

§ Violation check

slide-20
SLIDE 20

MMDIAG

Chunyi Peng HotMobile'16 20

MM Automata

Config. Collector Scenario Emulator static dynamic Instability Analyzer Counter examples Empirical Validation

§ Scenario reconstruction and experiments

§ Configurations and observations in counterexample § Trace collection and comparison

slide-21
SLIDE 21

Real-World Findings

  • One top-tier US carrier
  • Los Angeles and Columbus

63 locations (outdoor) 50 spots (indoor)

Chunyi Peng HotMobile'16 21

slide-22
SLIDE 22

Preliminary Results

§ 17 loops (idle) § 1 loop (active)

Chunyi Peng HotMobile'16 22

c1 (4G, band 17) c2 (4G, band 2) c3 (4G, band 4) c4 (Femtocell) c5 (3G, band 850) c6 (3G, band 1900) c7 (2G) L1: 4G-Femtoell-3G L2: 4G-Femtoell-2G-3G L3: 4G-4G

slide-23
SLIDE 23

Four Classes (Root Causes)

§ #1: uncoordinated handoff goals

¨ 8 variants,

4G-Femto-3G

Chunyi Peng HotMobile'16 23

c1 (4G, band 17) c2 (4G, band 2) c3 (4G, band 4) c4 (Femtocell) c5 (3G, band 850) c6 (3G, band 1900) c7 (2G) L1: 4G-Femtoell-3G L2: 4G-Femtoell-2G-3G L3: 4G-4G

slide-24
SLIDE 24

Four Classes

§ #2: device-side misconfiguration

¨ 8 variants, 4G-

Femto-2G-3G

Chunyi Peng HotMobile'16 24

c1 (4G, band 17) c2 (4G, band 2) c3 (4G, band 4) c4 (Femtocell) c5 (3G, band 850) c6 (3G, band 1900) c7 (2G) L1: 4G-Femtoell-3G L2: 4G-Femtoell-2G-3G L3: 4G-4G

slide-25
SLIDE 25

Four Classes

§ #3: Imprudent 4G upgrade

¨ One 4G-only

loop

§ #4: uncoordinated load balancing

¨ One 4G-only

loop (active)

Chunyi Peng HotMobile'16 25

c1 (4G, band 17) c2 (4G, band 2) c3 (4G, band 4) c4 (Femtocell) c5 (3G, band 850) c6 (3G, band 1900) c7 (2G) L1: 4G-Femtoell-3G L2: 4G-Femtoell-2G-3G L3: 4G-4G

slide-26
SLIDE 26

Takeaway

§ Largely stable in practice

¨ Instability mainly caused by Femtocells or incompatible

upgrades

§ But in principle, instability likely exists

¨ Distributed nature ¨ Diversity and external (non-carrier) factors in case of

heterogeneous networks (femtocells, small cells, WiFi, etc)

Chunyi Peng HotMobile'16 26

slide-27
SLIDE 27

Open Issues

§ Non-stability properties

¨ Handoff converges to an undesirable choice (3G/2G

when 4G available)

§ Cooperate with network-side efforts § From detection to fix

¨ Report identified problems to carriers ¨ Assist end-devices to intervene the loop Chunyi Peng HotMobile'16 27

slide-28
SLIDE 28

Summary

§ A first look at instability in mobility management

  • ver cellular network

§ Disclose real-world persistent loops caused by misconfigurations and policy conflicts § Propose MMDIAG to detect unstable MM § Call for more attention and efforts

Chunyi Peng HotMobile'16 28

slide-29
SLIDE 29

Thank you! Questions?

Chunyi Peng HotMobile'16 29