Upload
tyrone-mayer
View
21
Download
0
Embed Size (px)
DESCRIPTION
Secure Protocols for Supply Chain Collaboration: Capacity Allocation and Collaborative Forecasting. Vinayak Deshpande Krannert School of Management Purdue University Joint Work With: Mikhail Atallah Leroy B.Schwarz Keith Frikken. Our Goal. - PowerPoint PPT Presentation
Citation preview
1
Secure Protocols for Supply Chain Collaboration:
Capacity Allocation and Collaborative Forecasting
Vinayak Deshpande
Krannert School of ManagementPurdue University
Joint Work With:
Mikhail Atallah
Leroy B.Schwarz
Keith Frikken
2
Our Goal..
...we are developing protocols to enable Supply-Chain Partners to
Make Decisions that Cooperatively Achieve Desired System Goals without Revealing
Private Information
3
Two Supply Chain Problems..
• Capacity Allocation without revealing retailer’s order quantities
• Collaborative Forecasting without revealing private forecast information
4
A capacity allocation problem…
• Supplier Often Puts Customers “On Allocation”
- Each customer gets some or all of its order, based on metrics (e.g., Past Sales, Days-of-Supply)
• Examples:- Honda Odyssey- Flat Panel Monitors for PC’s
• Our Goal: Construct Secure Protocols for implementing the Optimal capacity allocation mechanisms
5
The Business Scenario
• Single supplier selling to N 2 independent retailers or manufacturers, whose orders are based on private demand information, and the supplier’s announced price and allocation policy
• Focus on situation when total orders exceed supplier’s capacity or inventory
6
......
Supplier
N RetailersRevenues Ri(qi,i)
Price & Quantity Allocation
{P(i), Q(i, -i)}
Model Framework
Capacity K at unit cost c
i – Retailer’s private information parameter
7
Linear Allocation Mechanism
Index the retailers in decreasing order of their order quantities,
i.e., q1 q2... qN. Retailer i is allocated Qi(q,n) where
ninqQ
niKqn
qnqQ
i
n
jiii
if 0),(
if )},0max(1
),(1
Where n is the largest integer such that Qi(q,n) 0 for all i.
Example: Retailers face newsvendor problem with normal demand distribution with mean , with an exponential prior on .
8
Proportional Allocation Mechanism
Retailer i is allocated Qi(q) where
}/,min{)(1
n
jiiii qKqqqQ
Example: Retailers face newsvendor problem with uniform demand distribution on [0, ] with Pareto prior
9
Can the linear and proportional allocation mechanisms be implemented without revealing the retailer order quantities (and hence the private information parameter ) to the supplier?
10
Secure Protocol for determining if capacity is tight
1. Run the secure summation protocol to compute X = R + i qi, where R is a random chosen by the
supplier, and X becomes known to the retailers not the supplier.
2. Supplier locally computes K+R3. Run a secure comparison protocol to determine
whether Y<X4. If answer is yes, capacity is tight.
Protocol determines if capacity is tight without revealing qi to the supplier, or K to the retailers.
11
Secure Proportional Allocation Protocol
1. N retailers cooperatively choose a large Random R’, not known to the supplier
2. Each retailer sends R’*qi to the supplier3. Supplier sends D’ = i qi*R’/K to every retailer4. Every retailer computes it’s allocation qi’ as follows
qi’ = R’*qi/D’ = qi* K/ i qi
12
SPAP: Who Knows What?
• Supplier capacity known only to the supplier• Retailer orders known only to individual retailers• Retailer’s common random number not revealed to the supplier• Individual retailer’s allocated quantity revealed to both the supplier and the individual retailer• Sum of individual retailer orders not revealed to anyone
13
Secure Linear Allocation Protocol
1. Every retailer marks himself as “active”. Let be the set of active retailers and n = | |
2. Repeat steps 2(a)-(d) till n ceases to changea. Every Retailer generates a random Ri. Let R = Ri; no
single party knows Rb. Using secure summation protocols compute n and
D =iPqi-K+R, such that D is known only to the supplierc. If n is the same as in previous iteration, go to step 3.d. Run a secure summation protocol to compute
(D/ n) – (R/n) = (iP qi – K)/n which is the pain per active retailer. If this pain exceeds a retailers order quantity, he marks himself “passive”
3. Every retailer computes it’s allocation equal to their order quantity minus the pain per active retailer
14
SLAP: Who Knows What?
• Supplier capacity known only to the supplier• Retailer orders known only to individual retailers• Pain per active retailer revealed only to retailers• Individual retailer’s allocated quantity revealed to both the supplier and the individual retailer• Number of active retailers known to everyone
15
Secure Collaborative Forecasting
• develop protocols for constructing a joint forecast without revealing the retailers or suppliers private information
16
Industry Backdrop
• Collaborative Planning, Forecasting, and Replenishment (CPFR), an initiative of the Voluntary Intra-Industry Collaboration Society (VICS)– buyer and supplier share inventory-status, forecast, and
event-oriented information and collaboratively make replenishment decisions
– pilot program between Wal-Mart and Warner-Lambert, called CFAR: (www.cpfr.org)
• Challenges to CPFR– fear that competitively-sensitive “private information”
will be compromised– Necessary to protect “sensitive” forecast information
such as sales promotions from “leaking”
17
Business Scenario
• A supply-chain with two players, a supplier selling to a retailer.
• The retailer and the supplier receive independent “signals” about future market demand– e.g., a retailer has private information about “promotions”
that he may be planning to run in the future which can affect his forecast of demand;
– the supplier can receive signals about overall “market trends”
• Incorporating these “signals” can improve forecast accuracy
• But.. “signal” information should be kept private
18
Demand Model
, ,1 1
T Tr s
t r t i s t i ti i
d
• dt – demand in period t (observed by the retailer only)
• t,ir – Retailer’s signal about period t observed in period t-i
(private information to the retailer)
• t,is –Supplier’s signal about period t observed in period t-i
(private information to the supplier)
• , r , s – unknown parameters to be estimated from past observations
19
Forecasting Process
• In each period t, estimate , r , s by regressing the observations dt versus the observed signals t,i
r and t,i
s
• For the forecast horizon (T periods) construct the forecast using the following equation:
, , 1,...,
T Tr s
j r sj i j ii j t i j t
d j t t T
23
Secure Demand Forecasting ProtocolInput: The supplier knows the j,i
s and the retailer knows the j,ir , for all j,
i such that j = t + 1, ..., t + T and i = j − t, ..., T. The parameters , r , s are available in additively split form.
Output: Both supplier and retailer learn the forecast dj for all j = t + 1, ..., t + T.
Protocol Steps: 1. For each j {t + 1, ..., t + T}, the supplier computes vj
s = j,i
s. This is a “local” computation, as the supplier has all the j,i
s values. The retailer similarly computes vj
r = j,i
r for all j {t + 1, ..., t + T}.2. For each j {t + 1, ..., t + T}, the supplier and retailer run a split
multiplication protocol twice, once to compute wrj = rvr
j and once to compute ws
j = svsj (both in split fashion).
3. For each j {t + 1, ..., t + T}, the supplier and retailer run a split addition protocol to compute µ+ wr
j+ wsj, which is equal to dj . They
exchange their shares of each dj so they both learn its value.
24
Protocol Implementation Issues:
Protocols are verifiable
• The Logic of the Protocol is Auditable– Logic of Source Code Can be Audited
• Outputs Can be Tested– Outputs Can be Verified Given Known Inputs
25
Protocol Implementation Issues:
Other Advantages
• Valuable even in Trusted e.g. (intra-corporate) interactions– “Defense in depth” ! – Systems are hacked into, break-ins occur,
viruses occur, spy-ware, bad insiders, etc– Liability Decreased
• “Don’t send me your data even if you trust me”
• Impact on Litigation and Insurance Rates
26
Future Work
• Protocols for other supply-chain applications– Price-Masking– Bullwhip Scenarios
• Protocol implementation issues– Collusion by a subset of parties– Intrusion detection– Incentive issues and mechanism design