Toward Cloud Manufacturing: A Decision Guidance Framework for

Markets of Virtual Things

Xu Han

a

and Alexander Brodsky

b

Department of Computer Science, George Mason University, 4400 University Drive, Fairfax, U.S.A.

Keywords:

Markets of Virtual Things, Decision Guidance, Formal Framework, Cloud Manufacturing.

Abstract:

In the value creation chain today, entrepreneurs have been faced with stiff hindrance in turning their innovative

consistent with the requirement specs, (2) are Pareto-optimal in terms of customer-facing metrics such as cost,

product desirable characteristics and delivery terms; and (3) that are optimal in terms of customer utility func-

tion that is expressed in terms of customer facing metrics. We also propose the design of a repository of virtual

things and their artifacts, to be used in support of the virtual things’ markets. The proposed markets of vir-

tual things can lead to democratizing innovation by allowing entrepreneurs without design and manufacturing

expertise to bring their ideas to markets quickly.

1 INTRODUCTION

In the value creation chain today, entrepreneurs have

been faced with stiff hindrance in turning their in-

novative ideas into marketable products due to the

manufacturing and testing, or (2) assemble a team of

experts who can help translate and convey the ideas

to the designers, developers, testers and manufactur-

ers. That is, entrepreneurs lack accessibility to man-

ufacturing capacity, predictability and agility to get

to market. Furthermore, the current approaches typ-

ically lead to sub-optimal outcomes due to informa-

tion loss during the ideation and communication pro-

cesses.

For manufacturers, due to a lack of market infor-

2009), (Yu et al., 2011), (LaToza et al., 2013), (Shin

et al., 2017)), analysis and optimization ((Egge et al.,

2013), (Shao et al., 2018)). Parametric design seems

to point out a promising direction as it offers the flex-

ibility to the entrepreneurs - setting and changing the

parameters enables high customizability, and a large

set of design alternatives can be generated for the en-

trepreneurs to choose from. Recently, a number of

start-ups, such as Xometry, Kerfed and Physna, have

taken important complimentary steps to simplify the

uct and process designs necessary for manufacturing.

For different products, there is no consistent way to

reuse the models of the parts that can be shared. Fur-

thermore, when optimization is deployed to paramet-

ric designs, it is typically done in product and process

silos, as opposed to optimizing holistically across the

value chain, thus missing out the opportunity to reach

Pareto-optimal solutions in creating the products.

In order to enable the entrepreneurs to conceive

their ideas in a non-technical language but to ex-

systems. Bridging this gap is exactly the focus of this

paper.

More specifically, in this paper we develop a for-

mal mathematical framework for markets of virtual

things: parameterized products and services that can

be searched, composed and optimized. The proposed

framework formalizes the notions of virtual product

and service design and customer-facing specs as well

as requirements’ specs. Based on these formal con-

cepts, the framework also formalizes the notions of

is expressed in terms of customer facing metrics. We

also propose the design of a repository of virtual

things and their artifacts, to be used in support of the

market.

The paper is organized as follows. In Section 2 we

give an overview of the idea and associated key con-

3 MATHEMATICAL

5. analyticModel - is a function

, . . . , p

) in D

, a vector

, . . . , m

. We will denote resulting

, i = 1, . . . , n, using the (.) dot

vp = (domain, parametersSchema, metricSchema,

) ∈ M

,

Note: CFS(vp) defines the set of all feasible met-

ric vectors {(m

, . . . , m

)|(m

) ∈ M

M

, . . . , m

)}

In order to enable the users to search for the VPs,

we introduce the VP Requirements spec. The purpose

of the Requirements spec is for the users to get con-

ditions for search as they put constraints on the VP

space.

Definition 3.4. VP (consumer) Requirements Spec

(or RS for short) is a tuple

RS =< domain, metricSchema,

where

1. ob jectiveSchema is a set

(b) O

2. objectives: M

de-

fines objectives as a function of metrics i.e.,

ob jectives(m

) is a vector of objective

constraints in the metric space

4. OC : O

constraints in the objective space

The user search should only render the results that

fall in the range of the constraints domains determined

by the product characteristics and specified by the

user. For example, a user searches for a bike with

weight not greater than 10 kg, and price not greater

defined by: C(m

) = MC(m

, . . . , m

OC(ob jectives(m

Note also that VP Requirements spec defines a set

of all feasible metric vectors R:

R = {(m

, . . . , m

) ∈

To search for a VP in the marketplace, the user

specifies the VP Requirements Spec (RS), and the

, . . . , p

, . . . , p

) in parametersSchema, let

) be designated as decision vari-

• Let U : O

• RS and CFS match (or CFS is feasible w.r.t. RS,

, M

), . . . , (m

) =

1, . . . , l)o

such that o

, . . . , v

P

template(p

) are optimal w.r.t. RS, VP, U

and FP if:

– (p

) ∈ argmax(p

, . . . , p

Claim (follows directly from definitions): Given

CFS derived from VP, RS, U and let (m

, . . . , m

AM(p

). Then:

, . . . , m

) are Pareto-optimal w.r.t. RS

and C FS

) is optimal w.r.t. RS, V P and U ⇐⇒

formation function of zero or more virtual things into

zero or more virtual things. Virtual service is also as-

sociated with an analytic model, which gives metrics

of interest and the feasibility Boolean value, for every

instantiation of service parameters, which include pa-

rameters of input and output virtual things, as well as

additional internal parameters.

For example, a supply service transforms zero in-

put virtual things into output things that correspond to

a planned purchase (bike supply). Its metrics may in-

an (virtual) input metal part into a processed (virtual)

part, after a series of drilling and milling operations

(turning steel material into a bike frame). Or, a trans-

portation service transforms (virtual) items at some

V S = (input, out put, internalParametersSchema,

3. internalParametersSchema - is a set

, D

), . . . , (p

)}

PS = {(v

)}

5. analyticModel - is a function

) ∈ V

, a vector

6. f easMetric - is a Boolean metric name C ∈

mm

D = {T, F} ∈ MS (T to indicate that parameter

instantiation is feasible, and F otherwise)

Definition 3.8. VS Consumer Spec (projection on

< domain, metricSchema, MFC >

of pairs, where for i = 1, . . . , k

- the domain of m

, . . . , v

), . . . , (mm

, MM

)}

utility function, where O

, . . . , O

from the objective schema

OC = {(o

, O

), . . . , (o

We say that:

or CFS and RS are mutually consistent) if

) is satisfiable,

) is the metric constraint of

) ∈ MM

MM

is Pareto-optimal w.r.t. to RS and CFS if:

– for objective vectors (o

) =

ob jective(m

) and (o

, . . . , o

ob jective(m

) the following

) ∈ MM

is optimal w.r.t. RS, CFS and U

if: (mm

, . . . , mm

subject to CC(mm

)

• Given a vector FP = (v

, . . . , v

) of fixed pa-

, . . . , v

CC(v

, . . . , v

, . . . , v

, . . . , v

, . . . , v

) that

ob jective(AM(p

)) and

(o

, . . . , o

) = ob jective(AM(v

(∀i = 1, . . . , l, o

and the corresponding product p =

, . . . , v

, . . . , v

, . . . , v

)) subject to

, . . . , v

), where MC is the metric con-

and connected seamlessly to the conceptual schema.

In this section we show the organization of the fun-

damental artifacts (virtual products, virtual services,

specs, analytical models, utility functions, etc.) and

we demonstrate the composition of a V-thing with the

example of a virtual bike product.