A Secure and Efficient Cloud Resource Allocation Scheme with Trust Evaluation Mechanism Based on Combinatorial Double Auction Source:KSII Transactions on Internet and Information Systems vol. 11, no. 9, Sep. 2017 Authors: Yun-Hao Xia, Han-Shu Hong, Guo-Feng Lin, Zhi-Xin Sun Speaker: Chit-Jie Chew Date: 7/12/2018
Outline Introduction Related works Proposed scheme Experimental result Conclusions
Introduction - Double Auction $30 $40 Provider 1 Consumer 1 Profit: $30 $35 $50 Provider 2 Consumer 2 Auctioneer $30 $20 Consumer 3 Provider 3
Introduction - Combinatorial Double Auction Case 1: Provider 1, 2, 3 Profit: $220 Case 3: Provider 1, 3 Profit: $250 Case 3: Provider 1, 3 Profit: $250 Case 2: Provider 1, 2 Profit: $70 Case 4: Provider 2, 3 Profit: $200 Case 2: Provider 1, 2 Case 4: Provider 2, 3 Case 1: Provider 1, 2, 3 Case 3: Provider 1, 3 1 2 $100 1 $200 Auctioneer Auctioneer Auctioneer Auctioneer -$300 -$250 -$250 -$400 1 1 2 2 3 5 3 4 5 2 6 5 Consumer 2 Consumer 2 Consumer 2 +$200 +$300 +$300 +$300 Consumer 1 -1 -2 -2 -1 -2 -1 -3 -3 -3 Provider 1 Consumer 1 -$50 -$100 $0 $50 1 1 2 1 3 1 1 3 1 3 2 2 1 2 $150 2 3 $300 Consumer 3 +$120 Consumer 1 Consumer 1 Consumer 1 +$200 +$200 +$200 -1 -1 -1 -1 -1 -2 -1 -1 -1 -1 -1 Profit: $70 1 1 1 2 1 1 2 Profit: Profit: $200 $250 $100 Provider 2 Consumer 2 +$120 -1 -2 Consumer 3 Auctioneer 1 2 $120 1 4 $150 Profit: $220 2 Consumer 3 Provider 3
Related Works – Genetic Algorithm(GA) ∆𝑓=250−200=50 ∆𝑓=200−200=0 P1 P2 P3 C1 C2 C3 ∆𝑓=−280− −50 =−230 Population number Chromosome Fitness Parents 0 1 1 0 0 1 0 1 1 0 0 1 0 1 1 0 0 1 0 1 1 1 1 0 0 1 1 1 1 0 -$180 -$180 1 5 $200 1 1 0 1 1 0 0 1 0 1 1 0 0 1 0 1 1 0 0 1 0 0 1 1 1 -$50 -$50 2 4 $520 -1 -3 -3 Crossover 0 1 1 1 0 0 0 1 1 1 0 0 0 1 1 1 0 0 0 1 1 1 0 0 0 1 1 1 0 0 0 1 1 1 0 0 -$200 2 4 -$100 2 4 1 0 1 0 0 1 1 0 1 0 0 1 1 0 1 0 0 1 1 0 1 1 1 0 1 0 1 1 1 0 -$130 1 5 $250 1 Mutation 0 1 1 1 0 1 0 1 1 1 0 1 0 1 1 1 1 0 0 1 1 1 1 0 0 1 1 1 1 0 $20 -1 4 $200 $200 1 1 1 1 0 0 1 1 1 1 0 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 0 1 1 1 0 1 0 1 1 1 0 -$280 -$280 1 3 6 $250 $250 1 New Parents 2 4 1 0 1 1 1 0 1 0 1 1 1 0 1 0 1 1 0 0 $250 $250 1 -$50 0 1 1 1 1 0 0 1 1 0 0 1 $200 1 -$180 1 5
Related Works – Simulated Annealing Algorithm(SA) P1 P2 P3 C1 C2 C3 Chromosome Fitness 4 1 Fitness $70 1 1 0 1 0 1 Chromosome 4 2 -$50 1 0 1 1 0 0 4 1 $70 1 1 0 1 0 1 1 $50 0 1 0 1 0 0 Variation of temperature ∆𝑓=70 −(−$50)= 120 ∆𝑓=50 −$70=− 20 𝑒𝑥𝑝 ∆𝑓 𝑇 =𝑒𝑥𝑝 120 100 = 3.32 𝑒𝑥𝑝 ∆𝑓 𝑇 =𝑒𝑥𝑝 −20 90 = 0.8 𝐼𝑓 𝑒𝑥𝑝 ∆𝑓 𝑇 >𝑟𝑎𝑛𝑑𝑜𝑚 0~1 : 𝐼𝑓 𝑒𝑥𝑝 ∆𝑓 𝑇 >𝑟𝑎𝑛𝑑𝑜𝑚 0~1 : accept to new generation accept to new generation 𝑇=𝑇×0.9=100×0.9=90 𝑇=𝑇×0.9=90×0.9=81
Proposed scheme – Model Based on Combinatorial Double Auction
Proposed scheme – The Pricing Determination Model Notations Sign Definition 𝑙𝑜𝑎𝑑 𝑡 Current system load 𝑢𝑠𝑒𝑟_𝑎𝑚𝑜𝑢𝑛𝑡 The amount of demand by buyers 𝑝𝑟𝑜𝑣𝑖𝑑𝑒_𝑎𝑚𝑜𝑢𝑛𝑡 The amount of resources supply 𝑖 Resource consumer 𝑗 Resource 𝜔 𝑗 The weight of each kind of resources 𝑎 𝑖𝑗 The amount of consumer 𝑖 requires type 𝑗 resource ℎ Resource provider 𝑎 ℎ𝑗 The amount of provider ℎ requires type 𝑗 resource 𝑡 Timestamp 𝑝 ℎ 𝑡 The price of provider at the timestamp 𝑡 𝑝 ℎ Auction price of provider ℎ 𝑝 𝑖 𝑡 The price of consumer at the timestamp 𝑡 𝑝 𝑖 Bidding price of user 𝑖 𝑙𝑜𝑎𝑑 𝑡 = 𝑢𝑠𝑒𝑟_𝑎𝑚𝑜𝑢𝑛𝑡 𝑝𝑟𝑜𝑣𝑖𝑑𝑒_𝑎𝑚𝑜𝑢𝑛𝑡 𝑢𝑠𝑒𝑟_𝑎𝑚𝑜𝑢𝑛𝑡= 𝑖=1 𝑁 𝑗=1 𝐾 | 𝜔 𝑗 𝑎 𝑖𝑗 | provide_𝑎𝑚𝑜𝑢𝑛𝑡= ℎ=1 𝑁 𝑗=1 𝐾 | 𝜔 𝑗 𝑎 ℎ𝑗 | If 𝑙𝑜𝑎𝑑 𝑡 >1 𝑝 ℎ 𝑡 = 𝑝 ℎ × 1+ 1 1+ 𝑙𝑜𝑎𝑑 𝑡 If 𝑙𝑜𝑎𝑑 𝑡 <1 𝑝 𝑖 𝑡 = 𝑝 𝑖 × 1− 1 1+ 𝑙𝑜𝑎𝑑 𝑡
Proposed scheme – The Pricing Determination Model 1 2 $100 provide_𝑎𝑚𝑜𝑢𝑛𝑡= ℎ=1 𝑁 𝑗=1 𝐾 | 𝜔 𝑗 𝑎 ℎ𝑗 | = ℎ=1 2 𝑗=1 3 | 𝜔 𝑗 𝑎 ℎ𝑗 | Provider 1 = 0.3×1+0.3×2+0.4×1 + 0.3×1+0.4×4 =3.2 1 4 $150 user_𝑎𝑚𝑜𝑢𝑛𝑡= 𝑖=1 𝑁 𝑗=1 𝐾 | 𝜔 𝑗 𝑎 ℎ𝑗 | = ℎ=1 2 𝑗=1 3 | 𝜔 𝑗 𝑎 𝑖𝑗 | = 0.3×1+0.3×1+0.4×1 + 0.3×2+0.4×3 Provider 2 =2.8 1 $240 𝑝 𝑖 𝑡 = 𝑝 𝑖 × 1− 1 1+ 𝑙𝑜𝑎𝑑 𝑡 𝑙𝑜𝑎𝑑 𝑡 = 𝑢𝑠𝑒𝑟_𝑎𝑚𝑜𝑢𝑛𝑡 𝑝𝑟𝑜𝑣𝑖𝑑𝑒_𝑎𝑚𝑜𝑢𝑛𝑡 If 𝑙𝑜𝑎𝑑 𝑡 <1, Consumer 1 𝑝 1 𝑡 =240× 1− 1 1+0.875 𝑝 1 𝑡 =300× 1− 1 1+0.875 = 2.8 3.2 2 3 $300 =0.875 =112 =140 Consumer 2
Proposed scheme – The Pricing Determination Model 1 2 $100 1 2 $100 Provider 1 Provider 1 1 3 5 -$250 +$240 2 4 -$10 +$300 +$290 1 3 5 -$250 +$112 2 4 -$138 +$140 +$2 1 4 $150 1 4 $150 Provider 2 Provider 2 1 $240 1 $112 Consumer 1 Consumer 1 2 3 $300 2 3 $140 Consumer 2 Consumer 2
Proposed scheme – SAGA ∆𝑓=250−200=50 ∆𝑓=200−200=0 P1 P2 P3 C1 C2 C3 ∆𝑓=−280− −50 =−230 Chromosome Fitness Parents 0 1 1 0 0 1 0 1 1 1 1 0 -$180 1 5 $200 1 1 0 1 1 0 0 1 0 1 1 0 0 1 0 0 1 1 1 -$50 -$50 2 4 $520 -1 -3 -3 Mutation 0 1 1 1 0 1 0 1 1 1 1 0 0 1 1 1 1 0 $20 -1 4 $200 $200 1 1 1 1 0 0 1 1 1 1 0 0 1 1 0 1 1 1 0 1 0 1 1 1 0 -$280 -$280 1 3 6 $250 $250 1 𝑒𝑥𝑝 ∆𝑓 𝑇 =𝑒𝑥𝑝 −230 100 = 0.1 𝐼𝑓 𝑒𝑥𝑝 ∆𝑓 𝑇 >𝑟𝑎𝑛𝑑𝑜𝑚 0~1 : SA 0.1>0.05 New Parents 1 0 1 1 1 0 1 0 1 1 1 0 1 1 1 0 0 1 $250 $250 1 -$280 1 3 6 0 1 1 1 1 0 1 0 1 1 0 0 $200 1 -$50 2 4
Experimental result – The simulation of the efficiency of SAGA Parameter Value Population number 500 Chromosome number 16 Crossover probability 0.7 Mutation probability 0.08 Number of population genetics 20 Variation of temperature T 𝑇=𝑇×0.95
Experimental result – The simulation of the efficiency of SAGA
The percentage of malicious nodes Experimental result – The simulation of the efficiency of trust model Parameter Value 𝛿 0.8 𝜗 0.2 𝛼 𝛽 The percentage of malicious nodes Providers : 𝟏𝟎% User : 𝟏𝟎% The number of malicious nodes 𝑵×𝟏𝟎+𝟏 𝒏=𝟎~𝟗 𝑵×𝟏𝟎+𝟖
Experimental result – The simulation of the efficiency of trust model The percentage of malicious nodes The average number of successful transactions Malicious SP Normal SP Malicious users Normal users 10% 6.8 619.9 5.3 620.1 20% 5.4 534.6 2.2 535.4 30% 5.6 492.7 2.8 493.9 40% 5.9 507.3 3.7 508.7 50% 2.6 267.5 2.3 268.1
Experimental result – The simulation of the efficiency of trust model Fig. 13. Transaction’s number of normal\malicious users and SP under different percentages
Conclusions Secure and efficient combinatorial double auction SAGA Trust evaluation model