Joint Resource Dimensioning and Placement for Dependable Virtualized Services in Mobile Edge Clouds
Abstract : Mobile edge computing (MEC) is an emerging architecture for accommodating latency sensitive virtualized services (VSs). Many of these VSs are expected to be safety critical, and will have some form of reliability requirements. In order to support provisioning reliability to such VSs in MEC in an efficient and confidentiality preserving manner, in this paper we consider the joint resource dimensioning and placement problem for VSs with diverse reliability requirements, with the objective of minimizing the energy consumption. We formulate the problem as an integer programming problem, and prove that it is NP-hard. We propose a two-step approximation algorithm with bounded approximation ratio based on Lagrangian relaxation. We benchmark our algorithm against two greedy algorithms in realistic scenarios. The results show that the proposed solution is computationally efficient, scalable and can provide up to 30% reduction in energy consumption compared to greedy algorithms.
? The estimation of the workload and communication patterns in IoT-Fog/Edge networks has only been explored a little due the high heterogeneity of co-existing devices.
? Contrarily to existing average traffic characteristics, building dynamic traffic profiles and prediction mechanisms will enable more accurate, adaptive, and successful data offloading and resource allocation mechanisms.
? Most of the existing models for describing the performance of IoT-enabled applications are empirical and usually focus on a specific performance metric.
? In DRUID-NET, we will rely on existing estimation methods, such as, to estimate the workload of hardware constrained devices.
? A major challenge in MEC is to decide which services each edge node should host in order to satisfy the user demand, which we refer to as the service placement problem.
? The challenge in our problem is that many standard techniques for approximation algorithms, such as those used in, are either not applicable or can only provide a bad approximation ratio.
? We formulate the general service placement (GSP) problem as described above, and convert it to an equivalent problem that we call service placement with set constraints (SPSC) which is easier to approximate.
? Therefore, we transform GSP to an equivalent SPSC problem, and propose approximation algorithms for SPSC.
• One proposed solution is to simply prune the elements comprising the deep neural network for the EI service (e.g., remove the number of neurons/units or entire layers).
• Another proposed idea is to consider a EI model’s architecture being split across different tiers of the MEC architecture (e.g., one half is run on the edge and the other on the central cloud).
• We compare our proposed algorithms (EGP and AGP) to the optimal solution. Due to the hardness of the PIES problem, we consider a validation case to demonstrate EGP performance relative to the optimal solution2 and AGP.
• We thoroughly evaluate the proposed algorithm for making placement and scheduling decisions in both synthetic and realworld scenarios against the optimal solution and some baselines.
? Due to these challenges, existing work on service placement often has limitations in terms of practicality and performance guarantee.
? We now present the overall algorithm for SPSC, which has better empirical performance than SA1 and SA2 alone.
? We evaluate the performance of the proposed final algorithm, RSA (Algorithm 7), via simulations.
? Due to the heterogeneity of edge node characteristics and user locations, the performance of MEC varies depending on where the service is hosted.
? However, a fundamental difference between data files and service programs is that data files can be partitioned in arbitrary ways without affecting the cache efficiency.
|