extend MOTON to support (signed) integer, fixed/floating-point arithmetic, noise generation methods, and differentially private mechanisms

[liangzhao-darmstadt]

Hi,

Following operations are added to MOTION dev:

• [x] SecureUnsignedInteger (up to 128-bit) to support operations such as modulo reduction.
• [x] Tests (test_unsigned_integer_operations_bgmw.cpp, test_unsigned_integer_operations_bmr.cpp) are passed.
• [x] SecureSignedInteger (up to 128-bit) to support various operations (except conversion between fixed-point and floating-point arithmetic).
• [x] Tests (test_signed_integer_operations_bgmw.cpp, test_signed_integer_operations_bmr.cpp) are passed.
• [x] Fix MuxGate for Boolean circuit evaluation.
• [x] (Boolean circuit) based floating-point arithmetic (SecureFloatingPointCircuitABY)
• [x] Tests (test_floating_point_circuit_ABY_bgmw.cpp, test_floating_point_circuit_ABY_bmr.cpp) are passed.
• [x] (Boolean circuit) based fixed-point arithmetic (SecureFixedPointCircuitCBMC)
• [x] Tests (test_fixed_point_circuit_CBMC_bgmw.cpp, test_fixed_point_circuit_CBMC_bmr.cpp) are passed.
• [x] Conversion operations between the above number representation forms.
• [x] Noise generation methods for geometric noise, (discrete) Laplace noise, (discrete) Gaussian noise.
• [x] Various differentially private mechanisms: (insecure/secure) Laplace mechanism, (insecure/secure) Gaussian mechanism, discrete Laplace mechanism, discrete Gaussian mechanism.
• [x] Tests for noise generation methods and differentially private mechanisms.
• [x] Benchmarks for all above.

Note:

• Implementation of (arithmetic GMW based) fixed/floating-point arithmetic and corresponding building blocks requires 128-bit shared bits (sb_provider.cpp), and 256-bit square pairs (sp_provider.cpp).
• My implementation (huge modification of MOTION) of arithmetic GMW based fixed/floating-point arithmetic can be found at Securely-Realizing-Output-Privacy-in-MPC-using-Differential-Privacy