Computer aided network planning and techno-economic...

Preview:

Citation preview

Mitcsenkov Attila (BME TMIT, High Speed Networks Laboratory) 21 October 2014

Computer aided network planning and techno-economic

assessment of next generation optical access networks

Complex network planning methodology and framework

Image: Sumitomo Electric Lightwave

• Optical fiber “approaches” the customers:

- FTTC FTTB FTTH

• A.) Point-to-point Dedicated optical fiber to all customers

• B.) Point-multipoint systems Shared optical “feeder network”

- Active (e.g. Active Ethernet)

- Passive (PON)

Next Generation Access (NGA) networks

• „History”: ABON, BPON

• Currently deployed: - EPON (USA & Asia) / GPON (EU)

• Recent „upgrade”: - NGPON-1 10G EPON / 10G GPON

• Near future: - NGPON-2 TWDM PON, 40G

• Future development: - WDM PON (“virtual point-to-point”)

- Colorless vs. Tunable: no clear winner yet…

Passive Optical Networks (PON)

• Next Generation Access networks

- Short / near term demand: 100 Mb/s – 1 Gb/s per customer

- The common phrase: optical access is a “must”

• „FTTH boom”:

- Global Industry Analysts Inc.: 183.9 million FTTH/B subscriptions worldwide by 2015

- Deutche Telekom “will invest almost 30 billion euros over the next three years in order to ensure “the future of telecommunications.” (2012 – FTTC VDSL, Fiber & LTE)

- …

• What does it cost?

• Who will pay for it?

• Does it make sense?

Network design – The Motivation

• Economic aspects:

- Cost estimation

- Return of Investment, Cash Flow, … Business Case Analysis

- “What do engineers now about business…””

• Technical aspects:

- Quality parameters (bandwidth, delay, reliability, …)

- Support network planning: what and where to deploy?

- “Never let the marketing guys decide…”

• Decision support:

- Lowest cost vs. future proof technologies (e.g. VDSL vs. WDM PON)

- Best fitting solutions for each service area

The “crystal ball” : Techno-economic analysis

• Some fundamental questions:

- What is the cost of deploying a GPON network in XY area?

- And what about Active Ethernet, VDSL or Point-to-Point?

- Which one is the best?

- (what does “best” mean?)

The beginning Once upon a time…

?

?

?

• How to estimate the costs?

• Who will trust our estimation?

• Why exactly *** € / $? Why not less/more?

• We did a research of the literature (not to reinvent the wheel…)

- Rules of thumb: average cost per customer (but we didn’t have data for Hungary…)

- Geometric models to estimate the physical infrastructure (but our areas were not that regular / homogeneous…)

„Think, think, think…”

Geometric models 1. Triangle Model

E

A

B

C

D

R

F

F

FF

F

F

F F

F F

F

FP1

FP2

végpont

Kábel típus #1

Kábel típus #5

Kábel típus #4

Kábel típus #3

Kábel típus #2

α

AB BC R / 3 cos( / 2)

2CD CE R / 6 1 8 sin ( / 2)

DF R 0,132 0,336 n

Analytic calculations

Geometric models 2. Simplified Street Length model

Analytic calculations

n houses in a row

l = distance

between two houses

The central office

at the center

This serves n2

customers

2I n (n 1) l (n 1) l (n 1) l

n 1

i 1

F 4 l min i,n i n i

N: buildings in each street

L: distance between neighboring buildings

CO – Central Office in the geometric median, serving n2 customers

• How to estimate the costs?

• Who will trust our estimation?

• Why exactly *** € / $? Why not less/more?

• We did a research of the literature (not to reinvent the wheel…)

• Investigate the map

- “How big” network?

- How complicated “field”?

- How to adapt the geometric models?

- Knowing the network, we could answer all questions…

Think, think, think…”

• High level network design

- Network connections, system design, location of equipment, …

• Detailed and accurate information about the necessary physical infrastructure

- Network equipment

- Cable plant

- Optical fiber needs

• Provide “Bill of Material”

• Techno-economic analysis:

- Reliable preliminary cost estimation

- Comparison of technologies

Solution: Strategic design

• Design “guesswork” :

- Human creativity

- Time consuming

- N-1 unnecessary designs!

• Computer aided network design:

- Time and cost savings

- Repetitious calculations with variable parameters

- Highly complex

- Data availability

- Computing capacity

How to provide the “strategic design”?

• Geographic Information System (GIS) data and digital maps

• Increased computational capacity

• Modern algorithms

Map based, automatic design of access networks

• The technical preconditions were not met earlier (calling for the use of geometric models)

Opportunities that made the computer able to deliver

Challenges, model, cost function and constraints

Network planning as an optimization problem

What do we expect? What is a “good” network design?

What do we expect? What is a “good” network design?

Which one is the “better” PON network? Why?

• How to translate the problem to mathematics?

- Modeling

- Cost function

- Constraints

• What is an optimal, or at least good topology?

- Economic aspects: what costs to consider?

• Complexity

- NP-hard (@see Travelling Salesman Problem)

- Large scale problem instances (10.000+ households)

Challenges

e 2

e3

e1 e4co

coKözpont (Central

Office, CO)

Elosztóhálózati

összeköttetés

Törzshálózati

összeköttetés

Gráfpont

Végpont

Elosztópont

Megengedett

elosztópont hely

Nem használt

hálózati szakasz

e 2

e3

e1 e4

Gráfél

Gráfpont

e 2

e3

e1 e4co

coKözpont (Central

Office, CO)

Gráfpont

Végpont

Megengedett

elosztóponti hely

Lehetséges

hálózati

összeköttetés

CO

Törzshálózat Elosztóhálózat

Központ

(Central

Office, CO)

Végpont/

előfizető

Elosztópont

Törzshálózat

Elosztóhálózat

Előfizetői

csoport

{i}

Modeling – translate it to the computer! Point-to-multipoint networks

• Flexible, accurate, parametric

• Contains all necessary data – and nothing else!

- Potential network links

- Distances

- Customer data

- Equipment locations

Network graph model

Goal: minimize costs

CAPEX Deployment cost

OPEX Operation,

Maintenance

Topology-dependent costs

(e.g. cable plant)

Topology-independent costs

(e.g. user modems)

Cable plant

- Two levels: deployment + fiber costs

- Stepwise: discrete increments

- Existing infrastructure (!)

Network equipment

- Central Office

- Distribution Units

- Customers

Relation between these:

- More distribution units less cable/fiber infrastructure

- Two extremities: point-to-point network (1 distribution units / ∞ distribution units)

Cost function „All inclusive”

C0

C0+n·Cv

Kábelhálózat

költsége

#szálszám

Építési

költségek

Telepítési

költésg

Szálköltség

} Kábelhálózat

létrehozása

• Constraint set:

- Distance limitations

- Complete access network reach

- Feeder network reach

- Distribution network reach

- Differential distance

- Capacity of distribution units

Physical and technical constraints

Központ

Végpont

Elosztópont

(Pl.. ETH Switch)

Törzshálózat

Elosztóhálózat

Előfizetői

csoport

{i}

{i}

L feedm

ax

L feedmax

Törzshálózat

hatósugara

Ldist max

Törzshálózat

hatósugara

Elosztóhálózat

hatósugara Ldist max

Elosztóhálózat

hatósugara

How to solve?

The challenge is given…

• Fundamentals of optimization: “No Free Lunch” Theorem

» Specialized algorithms needed

- Follow the designers’ way of thinking

- Identify technology specific constraints and costs

1. Subproblems and decomposition 2. Specialization

• Costs:

- Cable plant dominates

- Splitter costs: low

• Network reach: 20 km not decisive

• Primary goal: optimal clustering

• Solution:

- Maximize shared cable segments

- Clustering on the Shortest Path Tree

- Typically tree topologies…

Passive Optical Networks (GPON)

Központ

Végpont

Splitter

Törzshálózat

Elosztóhálózat

Előfizetői

csoport

{i}

{i}

Lm

ax

Lmax

Hálózat

hatósugara

GPON

10GPON

(WDM-PON)

Branch Contracting Algorithm

• Costs: - Distribution Unit: high CAPEX + OPEX

- Cable plant: high costs…

• Network reach: 20+3 km (not sharp)

• Main goal: - Minimize distribution units / groups

- Optimal clustering:

- No overlapping

- Nearby distribution unit

• Solution: - Iterative “bottom-up” clustering

- Increase group size if possible

- Avoid overlapping

Active Optical Networks (AETH)

Központ

Végpont

Elosztópont

(Pl.. ETH Switch)

Törzshálózat

Elosztóhálózat

Előfizetői

csoport

{i}

{i}

L feedm

ax

L feedmax

Törzshálózat

hatósugara

Ldist max

Törzshálózat

hatósugara

Elosztóhálózat

hatósugara Ldist max

Elosztóhálózat

hatósugara

Active

Ethernet

• Costs: - DSLAM: high CAPEX + OPEX

- Copper: almost for free

• Constraints: - Decisive: 300/800 m copper reach

• Goal: - Minimize DSLAMs, respect reach

constraints

• Solution: - Greedy algorithm

- Prioritize critical customer nodes

- Top-down clustering

Hybrid copper + fiber (VDSL)

Központ

Végpont

Elosztópont

(DSLAM)

Törzshálózat

(optika)

Elosztóhálózat

(réz)

Előfizetői

csoport

{i}

{i}L feed

max

L feedmax

Törzshálózat

hatósugara

Ldist m

ax

Optikai törzshálózat

hatósugara

Rézhálózat

hatósugara

Ldist max Elosztóhálózat

(réz)

hatósugara

DSL

• Costs:

- Cable plant costly

- Especially the deployment costs (with scarce networks)

• Goal:

- Minimize trenching / digging

• Solution:

- Steiner-tree problem

- Distance Network Heuristics (DNH):

- Graph-transformations + minimal cost spanning tree

Point-to-point optical networks

Központ

Végpont

Hozzűférési

hálózat

Lm

ax

Lmax

Hálózat

hatósugara

Pont-pont

optika

Computer aided optical access network design in practice

AccessPlan Framework

• Economic analysis

- Cost estimation (CAPEX)

- Business Case Analysis

- Compare technologies, decision support

• Geographic + Infrastructure Data

- Handling digital maps

- Existing infrastructure

AccessPlan Framework Modular setup

• Build network model

- Graph, nodes and edges

- Cost data

- Technical parameters and constraints

• Topology optimization

- Clustering, group formulation

- Distribution Unit placement

- Connection establishment

AccessPlan input data

Strategic network design

System design

Location of network

equipment

Bill of Material

(cable plant, equipment)

• Detailed input for further analysis and comparison

• Technical:

- Performance analysis

- Possible and guaranteed service levels

- Reliability, identification (protection) of critical nodes

• Economic:

- Cost estimation

- Business Case Analysis

• Does everything provided by the geometric models, but more accurate, detailed and reliable…

AccessPlan Results

Let’s see an example…

Case study

AccessPlan: select area

Example: Sashegy (Budapest)

Sashegy, Budapest Districts 11/12

Area of a Central Office 5 km2 / 4500 customers

• Inhomogeneous area

- Family, mid-size and large block houses

- Nature reserve area

- Cemetery

• Existing infrastructure

- Partly re-usable substructures

- Existing copper network

• Different cabling technologies – makes a cost difference

- Aerial cables and poles

- Existing substructures

- New trenching

Parameters of the case study

• 82 km street system

- 60 km existing infrastructure

- 2 km aerial network

- 20 km new trenching

• Graph model:

- 8000 nodes

- 8500 edges

• Map, street system & infrastructure

Phase #1: Process input data

• Customer database

• Demand points:

- 4239 households in

- 1079 buildings

• Graph model:

- 10 300 nodes

- 10 424 edges

Phase #1: Process input data

AccessPlan: planning rules and cost data

After a few minutes…

• Topology

• System design

• Location of network equipment

Phase #2: Topology design

• Technical & physical constraints

• Cost database (cost function)

Phase #3: Transform back onto the map

• Topology information, network characteristics

- System design

- Location of network equipment

- Fiber, cable and equipment needs

• Business Case Analysis

- CAPEX, OPEX, Cash Flow, Payback, etc.

• Techno-economic analysis, comparison

- GPON, 10GPON, Active Ethernet, VDSL, Point-to-point fiber, …

• Any custom reports and analytics

- Based on detailed topology information

- Upon request of the network operator

Phase #4: Reports and analyses

AccessPlan: plans and results

AccessPlan: analyses

0

0,5

1

1,5

2

2,5

3

3,5

P2P GPON AETH VDSL

CAPEX M EUR

Techno-economic analysis 1/4 Reports: CAPEX estimation

Techno-economic analysis 2/4 Reports: Cash Flow

GPO

N

Pont-

pont

Eth

ern

et

Éves Cash Flow

Éves Cash Flow

Akkumulált Cash Flow

Akkumulált Cash Flow

0

100

200

300

400

500

600

700

800

5 000 800 300

EU

R /

zta

rtá

s

Háztartás / km2

VDSL: 25 Mb/s vs. 50 Mb/s

VDSL25 VDSL50

0

100

200

300

400

500

600

700

800

900

1 000

5 000 800 300

EU

R /

zta

rtá

s

Háztartás / km2

GPON: 50 Mb/s vs. 100 Mb/s

GPON50 GPON100

Techno-economic analysis 3/4 Reports: CAPEX vs. bandwidth

Techno-economic analysis 4/4 Reports: cost per bit

0 Ft

200 Ft

400 Ft

600 Ft

800 Ft

1 000 Ft

1 200 Ft

1 400 Ft

P2P FTTH P2P FTTB 10GPONFTTH

10GPONFTTB

WDMPONFTTH

WDMPONFTTB

Ft / bit/sec

Agglomeráció 1 021 Ft 1 035 Ft 587 Ft 548 Ft 243 Ft 244 Ft

Kertváros 925 Ft 1 088 Ft 644 Ft 617 Ft 236 Ft 295 Ft

Belváros 1 039 Ft 1 260 Ft 788 Ft 717 Ft 270 Ft 347 Ft

1 Mb/s sávszélesség ára (Ft / Mb/s)

Contact

Attila Mitcsenkov mitcsenkov@tmit.bme.hu

AccessPlan Framework http://accessplan.tmit.bme.hu/

Recommended